talk-data.com

Google SRE NYC proudly announces our last Google SRE NYC Tech Talk for 2025.

This event is co-sponsored by sentry.io. Thank you Sentry for your partnership!

Let's farewell 2025 with three amazing interactive short talks on Site Reliability and DevOps topics! As always the event will include an opportunity to mingle with the speakers and attendees over some light snacks and beverages after the talks.

The Meetup will take place on Tuesday, 16th of December 2025 at 6:00 PM at our Chelsea Markets office in NYC. The doors will open at 5:30 pm. Pls RSVP only if you're able to attend in-person, there will be no live streaming.

When RSVP'ing to this event, please enter your full name exactly as it appears on your government issued ID. You will be required to present your ID at check in.

Agenda: Paul Jaffre - Senior Developer Experience Engineer\, sentry.io One Trace to Rule Them All: Unifying Sentry Errors with OpenTelemetry tracing SREs face the challenge of operating reliable observability infrastructure while avoiding vendor lock-in from proprietary APM (Application Performance Monitoring) solutions. OpenTelemetry has become the standard for instrumenting applications, allowing teams to collect traces, metrics, and logs. But raw telemetry data isn't enough. SREs need tools to visualize, debug, and respond to production incidents quickly. Sentry now supports OTLP, enabling teams to send OpenTelemetry data directly to Sentry for analysis. This talk covers how Sentry's OTLP support works in practice: connecting frontend and backend traces across services, correlating logs with distributed traces, and using tools to identify slow queries and performance bottlenecks. We'll discuss the practical benefits for SREs, like faster incident resolution, better cross-team debugging, and the flexibility to change observability backends without re-instrumenting code. Paul’s background spans engineering, product management, UX design, and open source. He has a soft spot for dev tools and loses sleep over making things easy to understand and use. Paul has a dynamic professional background, from strategy to stability. His time at Krossover Intelligence established a strong foundation by blending Product Management with hands-on development, and he later focused on core reliability at MakerBot, where he implemented automated end-to-end testing and drove performance improvements. He then extended this expertise in stability and scale at Cypress.io, where he served as a Developer Experience Engineer, focusing on improving workflow, contribution, and usability for their widely adopted open-source community.

Thiara Ortiz - Cloud Gaming SRE Manager\, Netflix Managing Black Box Systems SREs often face ambiguity when managing black box systems (LLMs, Games, Poorly Understood Dependencies). We will discuss how Netflix monitors service health as black boxes using multiple measurement techniques to understand system behavior, aligning with the need for robust observability tools. These strategies are crucial for system reliability and user experience. By proactively identifying and resolving issues, we ensure smoother playback experience and maintain user trust, even as the platform continues to evolve and gain maturity. The principles shared within this talk can be expanded to other applications such as AI reliability in data quality and model deployments.

Thiara has worked at some of the largest internet companies in the world, Meta and Netflix. During her time at Meta, Thiara found a passion for distributed systems and bringing new hardware into production. Always curious to explore new solutions to complex problems, Thiara developed Fleet Scanner, internally known as Lemonaid, to perform memory, compute, and storage benchmarks on each Meta server in production. This service runs on over 5 million servers and continues to be utilized at Meta. Since Meta, Thiara has been working at Netflix as a Senior CDN Reliability engineer, and now, Cloud Gaming SRE Manager. When incidents occur and Netflix's systems do not behave as expected, Thiara can be found working and engaging the necessary teams to remediate these issues.

Andrew Espira - Platform and Site Reliability Engineer\, Founding Engineer kustode ML-Powered Predictive SRE: Using Behavioral Signals to Prevent Cluster Inefficiencies Before They Impact Production SREs managing ML clusters often discover resource inefficiencies and queue bottlenecks only after they've impacted production services. This talk presents a machine learning approach to predict these issues before they occur, transforming SRE from reactive firefighting to proactive system optimization. We demonstrate how to build predictive models using production cluster traces that identify two critical failure modes: (1) GPU under-utilization relative to requested resources, and (2) abnormal queue wait times that indicate impending service degradation. The SRE practitioners will learn how to extract early warning indicators from standard cluster logs, build ML models that provide actionable confidence scores for operational decisions, and take practical steps to integrate predictive analytics into existing SRE toolchains to achieve 50%+ reduction in resource waste and queue-related incidents This talk bridges the gap between traditional SRE observability and modern predictive analytics, showing how teams can evolve from reactive monitoring to intelligent, forward-looking reliability engineering" Andrew has over 8 years of experience architecting and maintaining large-scale distributed systems. He is the Founding Engineer of Kustode (kustode.com), where he develops cutting-edge reliability and observability solutions for modern infrastructure in the Insurance and health care solutions space. Currently pursuing graduate studies in Data Science at Saint Peter's University, he specializes in the intersection of reliability engineering and artificial intelligence. His research focuses on applying machine learning to operational challenges, with publications in peer-reviewed venues including ScienceDirect. He's passionate about making complex systems more predictable and maintainable through data-driven approaches. When not optimizing cluster performance or building the next generation of observability tools, Andrew enjoys contributing to open-source projects and mentoring early-career engineers in the SRE community.

Our Tech Talks series are for professional development and networking: no recruiters, sales or press please! Google is committed to providing a harassment-free and inclusive conference experience for everyone, and all participants must follow our Event Community Guidelines. The event will be photographed and video recorded.

Event space is limited! A reservation is required to attend. Reserve your spot today and share the event details with your SRE/DevOps friends 🙂

Date and Time

Aug 28, 2025 at 10 AM Pacific

Location

Virtual - Register for the Zoom

Exploiting Vulnerabilities In CV Models Through Adversarial Attacks

As AI and computer vision models are leveraged more broadly in society, we should be better prepared for adversarial attacks by bad actors. In this talk, we'll cover some of the common methods for performing adversarial attacks on CV models. Adversarial attacks are deliberate attempts to deceive neural networks into generating incorrect predictions by making subtle alterations to the input data.

About the Speaker

Elisa Chen is a data scientist at Meta on the Ads AI Infra team with 5+ years of experience in the industry.

EffiDec3D: An Optimized Decoder for High-Performance and Efficient 3D Medical Image Segmentation

Recent 3D deep networks such as SwinUNETR, SwinUNETRv2, and 3D UX-Net have shown promising performance by leveraging self-attention and large-kernel convolutions to capture the volumetric context. However, their substantial computational requirements limit their use in real-time and resource-constrained environments.

In this paper, we propose EffiDec3D, an optimized 3D decoder that employs a channel reduction strategy across all decoder stages and removes the high-resolution layers when their contribution to segmentation quality is minimal. Our optimized EffiDec3D decoder achieves a 96.4% reduction in #Params and a 93.0% reduction in #FLOPs compared to the decoder of original 3D UX-Net. Our extensive experiments on 12 different medical imaging tasks confirm that EffiDec3D not only significantly reduces the computational demands, but also maintains a performance level comparable to original models, thus establishing a new standard for efficient 3D medical image segmentation.

About the Speaker

Md Mostafijur Rahman is a final-year Ph.D. candidate in Electrical and Computer Engineering at The University of Texas at Austin, advised by Dr. Radu Marculescu, where he builds efficient AI methods for biomedical imaging tasks such as segmentation, synthesis, and diagnosis. By uniting efficient architectures with data-efficient training, his work delivers robust and efficient clinically deployable imaging solutions.

What Makes a Good AV Dataset? Lessons from the Front Lines of Sensor Calibration and Projection

Getting autonomous vehicle data ready for real use, whether for training, simulation, or evaluation, isn’t just about collecting LIDAR and camera frames. It’s about making sure every point lands where it should, in the right frame, at the right time.

In this talk, we’ll break down what it actually takes to go from raw logs to a clean, usable AV dataset. We’ll walk through the practical process of validating transformations, aligning coordinate systems, checking intrinsics and extrinsics, and making sure your projected points actually show up on camera images. Along the way, we’ll share a checklist of common failure points and hard-won debugging tips.

Finally, we’ll show how doing this right unlocks downstream tools like Omniverse Nurec and Cosmos—enabling powerful workflows like digital reconstruction, simulation, and large-scale synthetic data generation

About the Speaker

Daniel Gural is a seasoned Machine Learning Engineer at Voxel51 with a strong passion for empowering Data Scientists and ML Engineers to unlock the full potential of their data.

Clustering in Computer Vision: From Theory to Applications

In today’s AI landscape, these techniques are crucial. Clustering methods help organize unstructured data into meaningful groups, aiding knowledge discovery, feature analysis, and retrieval-augmented generation. From k-means to DBSCAN and hierarchical approaches like FINCH, selecting the right method is key: including balancing scalability, managing noise sensitivity, and fitting computational demands. This presentation provides an in-depth exploration of the current state-of-the-art of clustering techniques with a strong focus on their applications within computer vision.

About the Speaker

Constantin Seibold leads research group on the development of machine learning methods in the diagnostic and interventional radiology department at the university hospital Heidelberg. His research aims to improve the daily life of both doctors and patients.

Date and Time

Aug 28, 2025 at 10 AM Pacific

Location

Virtual - Register for the Zoom

Exploiting Vulnerabilities In CV Models Through Adversarial Attacks

As AI and computer vision models are leveraged more broadly in society, we should be better prepared for adversarial attacks by bad actors. In this talk, we'll cover some of the common methods for performing adversarial attacks on CV models. Adversarial attacks are deliberate attempts to deceive neural networks into generating incorrect predictions by making subtle alterations to the input data.

About the Speaker

Elisa Chen is a data scientist at Meta on the Ads AI Infra team with 5+ years of experience in the industry.

EffiDec3D: An Optimized Decoder for High-Performance and Efficient 3D Medical Image Segmentation

Recent 3D deep networks such as SwinUNETR, SwinUNETRv2, and 3D UX-Net have shown promising performance by leveraging self-attention and large-kernel convolutions to capture the volumetric context. However, their substantial computational requirements limit their use in real-time and resource-constrained environments.

In this paper, we propose EffiDec3D, an optimized 3D decoder that employs a channel reduction strategy across all decoder stages and removes the high-resolution layers when their contribution to segmentation quality is minimal. Our optimized EffiDec3D decoder achieves a 96.4% reduction in #Params and a 93.0% reduction in #FLOPs compared to the decoder of original 3D UX-Net. Our extensive experiments on 12 different medical imaging tasks confirm that EffiDec3D not only significantly reduces the computational demands, but also maintains a performance level comparable to original models, thus establishing a new standard for efficient 3D medical image segmentation.

About the Speaker

Md Mostafijur Rahman is a final-year Ph.D. candidate in Electrical and Computer Engineering at The University of Texas at Austin, advised by Dr. Radu Marculescu, where he builds efficient AI methods for biomedical imaging tasks such as segmentation, synthesis, and diagnosis. By uniting efficient architectures with data-efficient training, his work delivers robust and efficient clinically deployable imaging solutions.

What Makes a Good AV Dataset? Lessons from the Front Lines of Sensor Calibration and Projection

Getting autonomous vehicle data ready for real use, whether for training, simulation, or evaluation, isn’t just about collecting LIDAR and camera frames. It’s about making sure every point lands where it should, in the right frame, at the right time.

In this talk, we’ll break down what it actually takes to go from raw logs to a clean, usable AV dataset. We’ll walk through the practical process of validating transformations, aligning coordinate systems, checking intrinsics and extrinsics, and making sure your projected points actually show up on camera images. Along the way, we’ll share a checklist of common failure points and hard-won debugging tips.

Finally, we’ll show how doing this right unlocks downstream tools like Omniverse Nurec and Cosmos—enabling powerful workflows like digital reconstruction, simulation, and large-scale synthetic data generation

About the Speaker

Daniel Gural is a seasoned Machine Learning Engineer at Voxel51 with a strong passion for empowering Data Scientists and ML Engineers to unlock the full potential of their data.

Clustering in Computer Vision: From Theory to Applications

In today’s AI landscape, these techniques are crucial. Clustering methods help organize unstructured data into meaningful groups, aiding knowledge discovery, feature analysis, and retrieval-augmented generation. From k-means to DBSCAN and hierarchical approaches like FINCH, selecting the right method is key: including balancing scalability, managing noise sensitivity, and fitting computational demands. This presentation provides an in-depth exploration of the current state-of-the-art of clustering techniques with a strong focus on their applications within computer vision.

About the Speaker

Constantin Seibold leads research group on the development of machine learning methods in the diagnostic and interventional radiology department at the university hospital Heidelberg. His research aims to improve the daily life of both doctors and patients.

Date and Time

Aug 28, 2025 at 10 AM Pacific

Location

Virtual - Register for the Zoom

Exploiting Vulnerabilities In CV Models Through Adversarial Attacks

As AI and computer vision models are leveraged more broadly in society, we should be better prepared for adversarial attacks by bad actors. In this talk, we'll cover some of the common methods for performing adversarial attacks on CV models. Adversarial attacks are deliberate attempts to deceive neural networks into generating incorrect predictions by making subtle alterations to the input data.

About the Speaker

Elisa Chen is a data scientist at Meta on the Ads AI Infra team with 5+ years of experience in the industry.

EffiDec3D: An Optimized Decoder for High-Performance and Efficient 3D Medical Image Segmentation

Recent 3D deep networks such as SwinUNETR, SwinUNETRv2, and 3D UX-Net have shown promising performance by leveraging self-attention and large-kernel convolutions to capture the volumetric context. However, their substantial computational requirements limit their use in real-time and resource-constrained environments.

In this paper, we propose EffiDec3D, an optimized 3D decoder that employs a channel reduction strategy across all decoder stages and removes the high-resolution layers when their contribution to segmentation quality is minimal. Our optimized EffiDec3D decoder achieves a 96.4% reduction in #Params and a 93.0% reduction in #FLOPs compared to the decoder of original 3D UX-Net. Our extensive experiments on 12 different medical imaging tasks confirm that EffiDec3D not only significantly reduces the computational demands, but also maintains a performance level comparable to original models, thus establishing a new standard for efficient 3D medical image segmentation.

About the Speaker

Md Mostafijur Rahman is a final-year Ph.D. candidate in Electrical and Computer Engineering at The University of Texas at Austin, advised by Dr. Radu Marculescu, where he builds efficient AI methods for biomedical imaging tasks such as segmentation, synthesis, and diagnosis. By uniting efficient architectures with data-efficient training, his work delivers robust and efficient clinically deployable imaging solutions.

What Makes a Good AV Dataset? Lessons from the Front Lines of Sensor Calibration and Projection

Getting autonomous vehicle data ready for real use, whether for training, simulation, or evaluation, isn’t just about collecting LIDAR and camera frames. It’s about making sure every point lands where it should, in the right frame, at the right time.

In this talk, we’ll break down what it actually takes to go from raw logs to a clean, usable AV dataset. We’ll walk through the practical process of validating transformations, aligning coordinate systems, checking intrinsics and extrinsics, and making sure your projected points actually show up on camera images. Along the way, we’ll share a checklist of common failure points and hard-won debugging tips.

Finally, we’ll show how doing this right unlocks downstream tools like Omniverse Nurec and Cosmos—enabling powerful workflows like digital reconstruction, simulation, and large-scale synthetic data generation

About the Speaker

Daniel Gural is a seasoned Machine Learning Engineer at Voxel51 with a strong passion for empowering Data Scientists and ML Engineers to unlock the full potential of their data.

Clustering in Computer Vision: From Theory to Applications

In today’s AI landscape, these techniques are crucial. Clustering methods help organize unstructured data into meaningful groups, aiding knowledge discovery, feature analysis, and retrieval-augmented generation. From k-means to DBSCAN and hierarchical approaches like FINCH, selecting the right method is key: including balancing scalability, managing noise sensitivity, and fitting computational demands. This presentation provides an in-depth exploration of the current state-of-the-art of clustering techniques with a strong focus on their applications within computer vision.

About the Speaker

Constantin Seibold leads research group on the development of machine learning methods in the diagnostic and interventional radiology department at the university hospital Heidelberg. His research aims to improve the daily life of both doctors and patients.

Date and Time

Aug 28, 2025 at 10 AM Pacific

Location

Virtual - Register for the Zoom

Exploiting Vulnerabilities In CV Models Through Adversarial Attacks

As AI and computer vision models are leveraged more broadly in society, we should be better prepared for adversarial attacks by bad actors. In this talk, we'll cover some of the common methods for performing adversarial attacks on CV models. Adversarial attacks are deliberate attempts to deceive neural networks into generating incorrect predictions by making subtle alterations to the input data.

About the Speaker

Elisa Chen is a data scientist at Meta on the Ads AI Infra team with 5+ years of experience in the industry.

EffiDec3D: An Optimized Decoder for High-Performance and Efficient 3D Medical Image Segmentation

Recent 3D deep networks such as SwinUNETR, SwinUNETRv2, and 3D UX-Net have shown promising performance by leveraging self-attention and large-kernel convolutions to capture the volumetric context. However, their substantial computational requirements limit their use in real-time and resource-constrained environments.

In this paper, we propose EffiDec3D, an optimized 3D decoder that employs a channel reduction strategy across all decoder stages and removes the high-resolution layers when their contribution to segmentation quality is minimal. Our optimized EffiDec3D decoder achieves a 96.4% reduction in #Params and a 93.0% reduction in #FLOPs compared to the decoder of original 3D UX-Net. Our extensive experiments on 12 different medical imaging tasks confirm that EffiDec3D not only significantly reduces the computational demands, but also maintains a performance level comparable to original models, thus establishing a new standard for efficient 3D medical image segmentation.

About the Speaker

Md Mostafijur Rahman is a final-year Ph.D. candidate in Electrical and Computer Engineering at The University of Texas at Austin, advised by Dr. Radu Marculescu, where he builds efficient AI methods for biomedical imaging tasks such as segmentation, synthesis, and diagnosis. By uniting efficient architectures with data-efficient training, his work delivers robust and efficient clinically deployable imaging solutions.

What Makes a Good AV Dataset? Lessons from the Front Lines of Sensor Calibration and Projection

Getting autonomous vehicle data ready for real use, whether for training, simulation, or evaluation, isn’t just about collecting LIDAR and camera frames. It’s about making sure every point lands where it should, in the right frame, at the right time.

In this talk, we’ll break down what it actually takes to go from raw logs to a clean, usable AV dataset. We’ll walk through the practical process of validating transformations, aligning coordinate systems, checking intrinsics and extrinsics, and making sure your projected points actually show up on camera images. Along the way, we’ll share a checklist of common failure points and hard-won debugging tips.

Finally, we’ll show how doing this right unlocks downstream tools like Omniverse Nurec and Cosmos—enabling powerful workflows like digital reconstruction, simulation, and large-scale synthetic data generation

About the Speaker

Daniel Gural is a seasoned Machine Learning Engineer at Voxel51 with a strong passion for empowering Data Scientists and ML Engineers to unlock the full potential of their data.

Clustering in Computer Vision: From Theory to Applications

In today’s AI landscape, these techniques are crucial. Clustering methods help organize unstructured data into meaningful groups, aiding knowledge discovery, feature analysis, and retrieval-augmented generation. From k-means to DBSCAN and hierarchical approaches like FINCH, selecting the right method is key: including balancing scalability, managing noise sensitivity, and fitting computational demands. This presentation provides an in-depth exploration of the current state-of-the-art of clustering techniques with a strong focus on their applications within computer vision.

About the Speaker

Constantin Seibold leads research group on the development of machine learning methods in the diagnostic and interventional radiology department at the university hospital Heidelberg. His research aims to improve the daily life of both doctors and patients.

Date and Time

Aug 28, 2025 at 10 AM Pacific

Location

Virtual - Register for the Zoom

Exploiting Vulnerabilities In CV Models Through Adversarial Attacks

As AI and computer vision models are leveraged more broadly in society, we should be better prepared for adversarial attacks by bad actors. In this talk, we'll cover some of the common methods for performing adversarial attacks on CV models. Adversarial attacks are deliberate attempts to deceive neural networks into generating incorrect predictions by making subtle alterations to the input data.

About the Speaker

Elisa Chen is a data scientist at Meta on the Ads AI Infra team with 5+ years of experience in the industry.

EffiDec3D: An Optimized Decoder for High-Performance and Efficient 3D Medical Image Segmentation

Recent 3D deep networks such as SwinUNETR, SwinUNETRv2, and 3D UX-Net have shown promising performance by leveraging self-attention and large-kernel convolutions to capture the volumetric context. However, their substantial computational requirements limit their use in real-time and resource-constrained environments.

In this paper, we propose EffiDec3D, an optimized 3D decoder that employs a channel reduction strategy across all decoder stages and removes the high-resolution layers when their contribution to segmentation quality is minimal. Our optimized EffiDec3D decoder achieves a 96.4% reduction in #Params and a 93.0% reduction in #FLOPs compared to the decoder of original 3D UX-Net. Our extensive experiments on 12 different medical imaging tasks confirm that EffiDec3D not only significantly reduces the computational demands, but also maintains a performance level comparable to original models, thus establishing a new standard for efficient 3D medical image segmentation.

About the Speaker

Md Mostafijur Rahman is a final-year Ph.D. candidate in Electrical and Computer Engineering at The University of Texas at Austin, advised by Dr. Radu Marculescu, where he builds efficient AI methods for biomedical imaging tasks such as segmentation, synthesis, and diagnosis. By uniting efficient architectures with data-efficient training, his work delivers robust and efficient clinically deployable imaging solutions.

What Makes a Good AV Dataset? Lessons from the Front Lines of Sensor Calibration and Projection

Getting autonomous vehicle data ready for real use, whether for training, simulation, or evaluation, isn’t just about collecting LIDAR and camera frames. It’s about making sure every point lands where it should, in the right frame, at the right time.

In this talk, we’ll break down what it actually takes to go from raw logs to a clean, usable AV dataset. We’ll walk through the practical process of validating transformations, aligning coordinate systems, checking intrinsics and extrinsics, and making sure your projected points actually show up on camera images. Along the way, we’ll share a checklist of common failure points and hard-won debugging tips.

Finally, we’ll show how doing this right unlocks downstream tools like Omniverse Nurec and Cosmos—enabling powerful workflows like digital reconstruction, simulation, and large-scale synthetic data generation

About the Speaker

Daniel Gural is a seasoned Machine Learning Engineer at Voxel51 with a strong passion for empowering Data Scientists and ML Engineers to unlock the full potential of their data.

Clustering in Computer Vision: From Theory to Applications

In today’s AI landscape, these techniques are crucial. Clustering methods help organize unstructured data into meaningful groups, aiding knowledge discovery, feature analysis, and retrieval-augmented generation. From k-means to DBSCAN and hierarchical approaches like FINCH, selecting the right method is key: including balancing scalability, managing noise sensitivity, and fitting computational demands. This presentation provides an in-depth exploration of the current state-of-the-art of clustering techniques with a strong focus on their applications within computer vision.

About the Speaker

Constantin Seibold leads research group on the development of machine learning methods in the diagnostic and interventional radiology department at the university hospital Heidelberg. His research aims to improve the daily life of both doctors and patients.

Date and Time

Aug 28, 2025 at 10 AM Pacific

Location

Virtual - Register for the Zoom

Exploiting Vulnerabilities In CV Models Through Adversarial Attacks

As AI and computer vision models are leveraged more broadly in society, we should be better prepared for adversarial attacks by bad actors. In this talk, we'll cover some of the common methods for performing adversarial attacks on CV models. Adversarial attacks are deliberate attempts to deceive neural networks into generating incorrect predictions by making subtle alterations to the input data.

About the Speaker

Elisa Chen is a data scientist at Meta on the Ads AI Infra team with 5+ years of experience in the industry.

EffiDec3D: An Optimized Decoder for High-Performance and Efficient 3D Medical Image Segmentation

Recent 3D deep networks such as SwinUNETR, SwinUNETRv2, and 3D UX-Net have shown promising performance by leveraging self-attention and large-kernel convolutions to capture the volumetric context. However, their substantial computational requirements limit their use in real-time and resource-constrained environments.

In this paper, we propose EffiDec3D, an optimized 3D decoder that employs a channel reduction strategy across all decoder stages and removes the high-resolution layers when their contribution to segmentation quality is minimal. Our optimized EffiDec3D decoder achieves a 96.4% reduction in #Params and a 93.0% reduction in #FLOPs compared to the decoder of original 3D UX-Net. Our extensive experiments on 12 different medical imaging tasks confirm that EffiDec3D not only significantly reduces the computational demands, but also maintains a performance level comparable to original models, thus establishing a new standard for efficient 3D medical image segmentation.

About the Speaker

Md Mostafijur Rahman is a final-year Ph.D. candidate in Electrical and Computer Engineering at The University of Texas at Austin, advised by Dr. Radu Marculescu, where he builds efficient AI methods for biomedical imaging tasks such as segmentation, synthesis, and diagnosis. By uniting efficient architectures with data-efficient training, his work delivers robust and efficient clinically deployable imaging solutions.

What Makes a Good AV Dataset? Lessons from the Front Lines of Sensor Calibration and Projection

Getting autonomous vehicle data ready for real use, whether for training, simulation, or evaluation, isn’t just about collecting LIDAR and camera frames. It’s about making sure every point lands where it should, in the right frame, at the right time.

In this talk, we’ll break down what it actually takes to go from raw logs to a clean, usable AV dataset. We’ll walk through the practical process of validating transformations, aligning coordinate systems, checking intrinsics and extrinsics, and making sure your projected points actually show up on camera images. Along the way, we’ll share a checklist of common failure points and hard-won debugging tips.

Finally, we’ll show how doing this right unlocks downstream tools like Omniverse Nurec and Cosmos—enabling powerful workflows like digital reconstruction, simulation, and large-scale synthetic data generation

About the Speaker

Daniel Gural is a seasoned Machine Learning Engineer at Voxel51 with a strong passion for empowering Data Scientists and ML Engineers to unlock the full potential of their data.

Clustering in Computer Vision: From Theory to Applications

In today’s AI landscape, these techniques are crucial. Clustering methods help organize unstructured data into meaningful groups, aiding knowledge discovery, feature analysis, and retrieval-augmented generation. From k-means to DBSCAN and hierarchical approaches like FINCH, selecting the right method is key: including balancing scalability, managing noise sensitivity, and fitting computational demands. This presentation provides an in-depth exploration of the current state-of-the-art of clustering techniques with a strong focus on their applications within computer vision.

About the Speaker

Constantin Seibold leads research group on the development of machine learning methods in the diagnostic and interventional radiology department at the university hospital Heidelberg. His research aims to improve the daily life of both doctors and patients.

Date and Time

Aug 28, 2025 at 10 AM Pacific

Location

Virtual - Register for the Zoom

Exploiting Vulnerabilities In CV Models Through Adversarial Attacks

As AI and computer vision models are leveraged more broadly in society, we should be better prepared for adversarial attacks by bad actors. In this talk, we'll cover some of the common methods for performing adversarial attacks on CV models. Adversarial attacks are deliberate attempts to deceive neural networks into generating incorrect predictions by making subtle alterations to the input data.

About the Speaker

Elisa Chen is a data scientist at Meta on the Ads AI Infra team with 5+ years of experience in the industry.

EffiDec3D: An Optimized Decoder for High-Performance and Efficient 3D Medical Image Segmentation

Recent 3D deep networks such as SwinUNETR, SwinUNETRv2, and 3D UX-Net have shown promising performance by leveraging self-attention and large-kernel convolutions to capture the volumetric context. However, their substantial computational requirements limit their use in real-time and resource-constrained environments.

In this paper, we propose EffiDec3D, an optimized 3D decoder that employs a channel reduction strategy across all decoder stages and removes the high-resolution layers when their contribution to segmentation quality is minimal. Our optimized EffiDec3D decoder achieves a 96.4% reduction in #Params and a 93.0% reduction in #FLOPs compared to the decoder of original 3D UX-Net. Our extensive experiments on 12 different medical imaging tasks confirm that EffiDec3D not only significantly reduces the computational demands, but also maintains a performance level comparable to original models, thus establishing a new standard for efficient 3D medical image segmentation.

About the Speaker

Md Mostafijur Rahman is a final-year Ph.D. candidate in Electrical and Computer Engineering at The University of Texas at Austin, advised by Dr. Radu Marculescu, where he builds efficient AI methods for biomedical imaging tasks such as segmentation, synthesis, and diagnosis. By uniting efficient architectures with data-efficient training, his work delivers robust and efficient clinically deployable imaging solutions.

What Makes a Good AV Dataset? Lessons from the Front Lines of Sensor Calibration and Projection

Getting autonomous vehicle data ready for real use, whether for training, simulation, or evaluation, isn’t just about collecting LIDAR and camera frames. It’s about making sure every point lands where it should, in the right frame, at the right time.

In this talk, we’ll break down what it actually takes to go from raw logs to a clean, usable AV dataset. We’ll walk through the practical process of validating transformations, aligning coordinate systems, checking intrinsics and extrinsics, and making sure your projected points actually show up on camera images. Along the way, we’ll share a checklist of common failure points and hard-won debugging tips.

Finally, we’ll show how doing this right unlocks downstream tools like Omniverse Nurec and Cosmos—enabling powerful workflows like digital reconstruction, simulation, and large-scale synthetic data generation

About the Speaker

Daniel Gural is a seasoned Machine Learning Engineer at Voxel51 with a strong passion for empowering Data Scientists and ML Engineers to unlock the full potential of their data.

Clustering in Computer Vision: From Theory to Applications

In today’s AI landscape, these techniques are crucial. Clustering methods help organize unstructured data into meaningful groups, aiding knowledge discovery, feature analysis, and retrieval-augmented generation. From k-means to DBSCAN and hierarchical approaches like FINCH, selecting the right method is key: including balancing scalability, managing noise sensitivity, and fitting computational demands. This presentation provides an in-depth exploration of the current state-of-the-art of clustering techniques with a strong focus on their applications within computer vision.

About the Speaker

Constantin Seibold leads research group on the development of machine learning methods in the diagnostic and interventional radiology department at the university hospital Heidelberg. His research aims to improve the daily life of both doctors and patients.

Register to reserve your spot

Date and Time

April 25, 2025 from 5:30 PM to 8:30 PM

Location

The Meetup will take place at MotionLab.Berlin, Bouchéstraße 12/Halle 20 in Berlin

Leaving No Pixels Behind: Deep Learning for Perfect Cutouts

Removing backgrounds from images is a challenging task, even for advanced deep learning models. The human eye is highly sensitive to minor imperfections, making high-quality outcomes crucial. In this talk, Imran Kocabiyik will demonstrate how withoutbg achieves clean, natural-looking image extractions while addressing the issues of costly training data and the need to handle diverse image types. Their approach effectively balances intelligent model design and meticulous data selection, resulting in impressive performance suited for real-world applications.

About the Speaker

Imran Kocabiyik is a technologist building AI-driven tools for marketing and creative automation. He is the founder of withoutbg and a Senior Data Scientist at Klarna.

AI on the Dance Floor: Multimodal Segmentation of Choreography Videos

Ever struggled to learn a dance routine by constantly rewinding YouTube videos? In this talk, Paras presents an approach based on temporal convolutional networks and pose estimation to automatically segment choreography videos into individual moves by leveraging both audio and visual modalities.

About the Speaker

Dr. Paras Mehta is a computer scientist and co-founder of sylby, where he leads AI engineering for pronunciation training and language learning applications.

When Images Look Alike: Intro to Dataset Curation

This talk introduces dataset curation in computer vision, focusing on visually similar images. We discuss use cases in vacation rental search and art recommendations. We demonstrate how Voxel51 helps identify image similarity, improving data quality and model reliability.

About the Speaker

Antonio Rueda-Toicen, an AI Engineer in Berlin, has extensive experience in deploying machine learning models and has taught over 300 professionals. He is currently a Research Scientist at the Hasso Plattner Institute. Since 2019, he has organized the Berlin Computer Vision Group and taught at Berlin’s Data Science Retreat. He specializes in computer vision, cloud technologies, and machine learning. Antonio is also a certified instructor of deep learning and diffusion models in NVIDIA’s Deep Learning Institute.

EnvisionHGdetector: A Framework for Detecting and Analyzing Hand Gestures During Speech

We present EnvisionHGdetector, a toolkit for studying hand movements during speech. It measures hand motion, compares gestures, and labels gesture segments using Mediapipe tracking and a custom neural network. Tested on over 8,000 gestures, it achieved approximately 75% accuracy. We also discuss plans to improve accessibility for gesture researchers.

About the Speaker

Sharjeel Shaikh is currently pursuing an MSc in Data Science at the University of Potsdam. He works at HPI on Gesture Detection and Data Masking.

Our friends at Floryn are hosting the upcoming event at their office in Den Bosch (5 min walk from train station).

17:30 – 🍕 Welcome

18:00 – 🎤 Lights, dbt, Action: Making Analytics Engineering Visible (and Fast) Annebelle Olminkhof (Data Analyst) & Tijs Bronnenberg (Business Analyst) @ Floryn

18:30 – 🎤 Data analyst AI Agent powered by dbt Metadata - Daniel Herrera (Analytics Engineer & Developer Advocate) @ Teradata

19:00 –🥤 Drinks & Snacks

--- About the talks

🎤 Lights, dbt, Action: Making Analytics Engineering Visible (and Fast)

In 2022, our team of three data analysts at Floryn implemented dbt to build a more scalable and structured analytics workflow. At the time, most of our business logic was embedded in LookML within Looker, and dbt was more of a “nice to have” than a core component of our workflow. That changed last year when we migrated to a new BI tool, forcing us to extract all our LookML-based transformations into dbt.

This transition made us realize how much of our logic had been siloed within Looker, and it became the catalyst for fully centralizing our data models in dbt. By making dbt the foundation of our data analytics products, we standardized data transformation, improved data quality, and created a more scalable approach to managing our data.

Beyond improving our data models, dbt has enabled us to develop entirely new analytics products that wouldn’t have been possible before. With dbt as our single source of truth, our analytics engineers can now build cleaner, more reliable models while ensuring consistency across all reporting and analysis. We’ve leveraged dbt to develop metric trees that provide deeper insights into business performance as well as a data-driven warning system. By making dbt central to our analytics strategy, we’ve enhanced trust in our data and unlocked new opportunities for delivering meaningful insights.

In this talk, we'll share our journey from LookML-dependent modeling to a fully dbt-driven analytics framework, the challenges we faced, and the lessons we learned along the way. Whether you're considering dbt for your organization or looking to scale your analytics capabilities, our story highlights the power of a well-structured, centralized data strategy

🎤 Data analyst AI Agent powered by dbt Metadata

Generative AI adjacent concepts terms like "Agentic AI" or "vibe coding," are frequently used or misused as marketing hooks rather than as practical frameworks for understanding the technical reality of generative AI.

In this talk, we aim to cut through the noise by building a data analyst AI agent completely from scratch. We will not rely on any libraries or frameworks. Instead, we will focus on what it actually takes to create an agent that can generate insights from data.

One of the biggest challenges when working with large language models for sql query generation is providing the right context. Helping the model understand the structure and meaning of your data — including databases, tables, and columns and what they contain — is often the hardest part. However, if you are using a dbt, you already have access to rich metadata that can be passed to the model. This gives it the necessary understanding of the available data structures to generate accurate queries.

In this session, we will walk through how to use that metadata effectively and how to connect an LLM to your database. By the end, you’ll have a clear understanding of what it takes to build a functioning data analyst agent from the ground up.

--- Join the dbt Slack community: https://community.getdbt.com/ Join the conversation in the #local-netherlands channel in dbt Slack to connect with other data practitioners locally.

To attend, please read the Required Participation Language for In-Person Events with dbt Labs: https://www.getdbt.com/legal/health-and-safety-policy

Let’s kick things off for another meetup, this time focussing on scaling analytics with dbt and simplifying Kubernetes job execution. Join us for an engaging meetup on March 13th in Berlin!

Scaling Analytics with dbt: From Implementation to Impact

Managing data transformations, ensuring data quality, and fostering collaboration across teams present significant challenges in data management. This talk will explore the implementation of dbt in a project, highlighting its impact on workflow efficiency, key benefits, and encountered challenges.

Kube API: Simplifying Kubernetes Job Execution

Let’s focus on how Kube API simplifies and streamlines interactions with Kubernetes, particularly in managing jobs and workloads through a unified interface. We will share the concept of Kube API as a custom wrapper built on top of Kubernetes, acting as a microservice that provides a seamless experience for clients. The idea is to share how the REST API makes it easy for any one to create, manage, and monitor Kubernetes jobs without having to deal with Kubernetes’ inherent complexities.

What to expect:

• Expert talks
• Interactive Q&A
• Networking opportunities
• Some snacks & drinks :)

Timetable:

• 18:30 - Event admission
• 18:50 - Welcome & Introduction
• 19:00 - Scaling Analytics with dbt (Ceren Gündoğdu)
• 19:30 - 5 minutes break
• 19:35 - Simplifying Kubernetes Job Execution (Ajit Gosavi)
• 20:05 - Snacks, Drinks & Networking
• 21:30 - End

More on the -> applydata data engineering meetup page.

Our goal is to form a local data-loving community, so join us and let's talk Data together!

---

At the event, sound, image and video recordings are created and published for documentation purposes as well as for the presentation of the event in publicly accessible media, on websites and blogs and for presentation on social media. By participating the event, the participant implicitly consents to the aforementioned photo and/or video recordings. Find more information on data protection here.

Register for the event to reserve your spot!

Date and Time Feb 7, 2025 from 5:30 PM to 8:30 PM

Location The Meetup will take place at MotionLab.Berlin, Bouchéstraße 12/Halle 20 in Berlin

Smart Data Loops: A New Paradigm for AI Development and Anomaly Detection

In the era of autonomous driving, the quality and efficiency of AI development hinge on the ability to manage data intelligently. This talk introduces the concept of Smart Data Loop, a novel paradigm that revolutionizes data handling by improving out-of-distribution detection, and leveraging trigger functions to refine AI models continuously. We will explore how these innovative approaches enhance anomaly detection and streamline AI workflows.

About the Speaker

Dr. Azarm Nowzad holds a PhD in Computer Science and serves as the Technical Project Lead and Product Owner for “Data for AI” at Continental Automotive. She is currently leading the publicly funded project “justbetterDATA”, which focuses on developing efficient and highly accurate data generation methods for AI applications, particularly in the field of autonomous driving. With her expertise in computer vision and AI, she plays a pivotal role in advancing data-driven solutions for next-generation mobility.

All About Agentic AI

Today, the concept of Agentic AI is shaping how we think about intelligent systems. These are AI systems designed to act autonomously, making decisions, completing tasks, and interacting with their environment—beyond traditional AI models. Understanding how to design and develop Agentic AI products is essential for staying ahead in the competitive landscape of AI-driven innovation. In this talk, Dr. Arman Nassirtoussi introduces Agentic AI. He’ll cover how these systems differ from standard AI, the evolving architectures that support them, and why they’re becoming critical.

About the Speaker

Dr. Arman Nassirtoussi earned his PhD in AI over a decade ago, focusing on predictive AI algorithms for intraday financial trading using Natural Language Processing (NLP), sentiment analysis, and text mining of online news. His main publication has quickly received over 1,200 citations on Google Scholar. Arman has led large data engineering, data science, and AI teams at companies like Henkel, Zalando, and T-Systems, helping build infrastructure, platforms, and products with a major focus on personalization and product analytics in e-commerce. Arman has also created a number of startups in multiple countries, and he is currently shaping a new one in the Agentic AI space.

Bridging Minds and Machines: Aligning Human Behavior and Machine Algorithm

As AI systems increasingly support human decision-making, integrating human-centered design principles into ML engineering has become essential. This talk bridges the foundational concepts of Human-Computer Interaction (HCI) with the complex demands of algorithmic decision-making, focusing on bidirectional Human-AI alignment, trust calibration, and Reciprocal Human-Machine Learning (RHML).

We explore the necessity of embedding human behavior and neurocognitive feedback loops into ML pipelines to enable adaptive and trustworthy systems. Addressing overtrust, undertrust, and trust miscalibration, we emphasize aligning ML systems with both high-performance metrics and user behavior, ensuring systems are effective and ethically aligned.

About the Speaker

Anke Borchers is an AI Strategist and Consultant specializing in Machine Learning (ML), Generative AI, and Trustworthy AI. With a background in Industrial and Communication Design and over 15 years of experience in innovation and business strategy, she bridges the gap between human-centered design and advanced AI systems.

Dedicated to crafting tailored solutions for the medical and business sectors, Anke highlights the critical importance of human-centered AI systems. She offers deep expertise in cognitive and machine decision-making, as well as AI Alignment, empowering organizations to develop AI solutions that are high-performing, ethically sound, and optimized to address user needs effectively.

Attention is All We Need: Using Transformers in Vision Tasks

Attention mechanism, initially developed for natural language processing, is now being effectively applied in Computer Vision. This talk will focus on how attention enables Visual Transformers to capture context and why they are overpowering the classical approaches to vision tasks.

About the Speaker

Kira Kravets is a Machine Learning engineer at Kertos, specializing in LLMs and the development of trustworthy AI systems. With experience in Computer Vision, particularly in the highly demanding medical field, she is passionate about building real-world AI applications with all the limitations and restrictions of production environments.

🎙️ Speaker: William Dean\, Corey Abshire\, Thomas Wiecki \| ⏰ Time: 5 PM UTC / 9 AM PT / 12 PM ET / 6 PM Berlin

In this event, we will discuss how customers can use Databricks to develop and productionize MMM models for their companies. By combining Databricks capabilities in consolidating, organizing and securing data pipelines and manage ML models and pipelines with PyMC-Marketing’s easy to use modelling capabilities, companies can bring develop sophisticated MMM models to help understand, optimize and forecast their marketing budgets.

📜 Outline of Talk / Agenda:

• 5 min: Intro to PyMC Labs and speakers
• 45 min: Presentation, panel discussion
• 10 min: Q&A

💼 About the speaker:

1. Corey Abshire, Senior AI Specialist Solutions Architect, DataBricks Corey Abshire is a Senior Specialist Solutions Architect focused on GenAI and ML in Communications, Media and Entertainment. Prior to Databricks, Corey was a Principal Machine Learning Engineer at Cummins Inc., working on critical data science and machine learning initiatives for quality, engineering, and manufacturing. He holds an M.S. in data science from Indiana University in Bloomington, IN.

🔗 Connect with Corey: 👉 Linkedin: https://www.linkedin.com/in/coreyabshire/

1. Will Dean, Principal Data Scientist, PyMC Labs Will Dean is a Statistician and Data Scientist with experience in geospatial and user analytics. He is passionate about Bayesian methods and using data visualization to tell a story. He is interested in software design and how it can make data problems easier and more enjoyable to solve.

🔗 Connect with Will Dean: 👉 Linkedin: https://www.linkedin.com/in/williambdean/ 👉 Github: https://github.com/wd60622/

💼 About the Host:

1. Thomas Wiecki (Founder of PyMC Labs) Dr. Thomas Wiecki is an author of PyMC, the leading platform for statistical data science. To help businesses solve some of their trickiest data science problems, he assembled a world-class team of Bayesian modelers and founded PyMC Labs -- the Bayesian consultancy. He did his PhD at Brown University studying cognitive neuroscience. 🔗 Connect with Thomas: 👉 Linkedin: https://www.linkedin.com/in/twiecki/ 👉 Website: https://www.pymc-labs.com/, https://twiecki.io/ 👉 GitHub: https://github.com/twiecki 👉 Twitter: https://twitter.com/twiecki

📖 Code of Conduct: Please note that participants are expected to abide by PyMC's Code of Conduct.

🔗 Connecting with PyMC Labs: 🌐 Website: https://www.pymc-labs.com/ 👥 LinkedIn: https://www.linkedin.com/company/pymc-labs/ 🐦 Twitter: https://twitter.com/pymc_labs 🎥 YouTube: https://www.youtube.com/c/PyMCLabs 🤝 Meetup: https://www.meetup.com/pymc-labs-online-meetup/

DevFest Berlin is back! This year back to Humboldt University of Berlin, with more than 25 talks & workshops, you can expect a whole day of learning, socialising, and engaging with a vibrant Berlin Tech community!

🎫 Get you ticket here: pretix.eu/devfestberlin/2024/ 🖍 Call for Papers still open: pretalx.com/devfest-berlin-2024/cfp

Agenda

Day 1

9:00 AM: Registration & Coffee 🥐 ☕️

9:45 AM: 🎤 Welcoming

10:00 AM: 🎤 Katya Vinnichenko - Introduction to Google Principles of Responsible AI

This year's DevFest explores how AI can improve lives globally, from business to healthcare to education. At Google we acknowledge AI's potential, while also recognising the challenges it presents. Thus, we are committed to helping you build and use AI responsibly, ensuring fairness and ethical practices.

In my talk you will learn: the main principles of responsible AI at Google; the ethical implications of AI; best practices for developing AI systems and integrating AI into Google products and services; last but not least – how AI will change the role of the developer as we know it.

10:50 AM: 🎤 Oleksii Antypov - DMARC Demystified

Discover the essential framework behind DMARC and how it secures email communication across the internet. This session covers the historical evolution of email security, dives into the common challenges of implementing DMARC, and provides actionable best practices for protecting your domain. Ideal for developers, security professionals, and anyone interested in safe email practices.

In a world where phishing and email spoofing are constant threats, DMARC stands as a vital defense mechanism. “DMARC Demystified” takes you through a journey from the origins of email security to the modern challenges and solutions that DMARC offers. We'll explore how DMARC works with SPF and DKIM, why it’s essential for organizations of all sizes, and the practical steps to ensure smooth implementation.

Expect an interactive timeline tracing the milestones of email security, detailed breakdowns of real-world cases, and insights into optimizing DMARC. Walk away with a deeper understanding of email protection, armed with knowledge to strengthen your email systems and protect against threats.

11:40 AM: 🎤 Marcin Chudy - Demystifying App Architecture: The LeanCode Guide

At LeanCode we developed over 40 Flutter apps, spanning from huge enterprise apps to nimble startup ventures. Some were developed by a single Flutter dev, some came into light through collaborative efforts across multiple teams. Each of them was different. Each of them presented unique challenges and taught us invaluable lessons.

In this talk, we invite you to explore different approaches to architecting Flutter apps. Central to our narrative will be the concept of architectural drivers - key factors or priorities that steer our decisions about how the app is structured and designed. We'll show how we leverage our experience when approaching new projects. Drawing from our successes and failures, we'll present our current Flutter stack which enables us to craft robust, scalable, and maintainable applications. While there is no silver bullet for Flutter architecture, we can still have some sensible defaults.

Why do we use BLoC for state management? Why not Riverpod? Why do we love hook

12:30 PM: 🎤 Danny Preussler - Ten things you heard about testing that might be wrong

Testing became an essential part of Android development. Many conference talks have been given and even more best practices have been written.

But what if, as time evolved, some of the things we thought were true, changed?

Let’s start questioning some of these in this talk: Are flaky tests fixable? Are mocks even harmful? Is DI about testing? Did we understand testing in isolation properly? Is the test pyramid still valid? And in times of AI, should we generate tests?

Come and join my session to learn more!

1:10 PM: Lunch 🍔🥤

2:40 PM: 🎤 Andrey Sitnik - Privacy-first architecture: alternatives to GDPR popup and local-first

Why and how modern developers could increase the privacy of modern Web.

The popularity of clouds, the rise of huge monopolies across the internet, and the growth of shady data brokers recently have made the world a much more dangerous place for ordinary people—here is how we fix it.

In this talk, Andrey Sitnik, the creator of PostCSS and the privacy-first open-source RSS reader, will explain how we can stop this dangerous trend and make the web a private place again. — Beginners will find simple steps, which can be applied to any website — Advanced developers will get practical insights into new local-first architecture — Privacy experts could find useful unique privacy tricks from a global world perspective and beyond just U.S. privacy risks

3:30 PM: 🎤 Raphaël VO - Largest Contentful Paint - The unheard story

Largest Contentful Paint (LCP) is more than a speed metric — it's the unseen factor shaping user experiences and impacting SEO. While often overlooked, LCP reveals when a page’s core content is truly ready, affecting how users perceive load time and usability. This talk uncovers LCP’s role, why it matters more than we think, and simple strategies to boost LCP for better engagement and rankings. Discover the hidden story behind one of web performance’s most crucial, yet understated metrics.

Did you know the speed of a single webpage element could decide if users stay or leave? Largest Contentful Paint (LCP) is that hidden hero, quietly working to load the most important content quickly. This talk unveils LCP’s role in creating faster, more engaging web experiences and why it’s key to winning user loyalty. Dive into the “unheard story” of LCP and discover practical tips to make your site not only faster but unforgettable.

4:20 PM: 🎤 Ash Davies - Navigation in a Multiplatform World: Choosing the Right Framework for your App

Navigation in mobile, desktop, and web applications is such a fundamental part of how we structure our architecture. In order to both obtain functional clarity, and abstraction from platform level implementation.

For a long time, there have been options available specific to each platform, and even options part of the platform framework itself. Though it can be difficult to find the right option for platform-agnostic code, ensuring consistency. Some go one step further, providing an opinionated guide on how to architecture your application.

In this talk, I'll evaluate the options available, how they differ, and to what type of applications they are best suited. Including how to get started with them, and the best practice guidelines on how to get the most out of them, for your application.

5:10 PM: 🎤 Vadim Makeev - You don’t know MathML. Almost nobody does

Do you speak math? Me neither. Still, math formulas have always been around: from Wikipedia articles to JavaScript APIs and even CSS docs. It looks so alien that I never had a clue how to express it on the web. Apparently, there’s a markup language for that. HTML for content, SVG for vector graphics, and MathML for math! And it’s pretty cross-browser, too. Let’s dive into the basics and quirks of the language of the universe. Even if math is not your love language, you might learn something interesting about the web platform.

Day 2

9:00 AM: Registration & Coffee 🥐 ☕️

10:00 AM: 🎤 Alex Mir – Accessibility matters

The regulators are here and now businesses will care about the a11y. Let's make the a11y compliance not just a formal check. I believe that it is our job as industry experts to understand why it is important and get our products ready for all groups of people.

10:50 AM: 🎤 Marco Gomiero - From Android to Multiplatform and beyond

With Kotlin Multiplatform getting increasingly established, many Android libraries became multiplatform.

But how to make an existing Android library multiplatform?

In this talk, we will cover the common challenges faced while migrating Android libraries to Kotlin Multiplatform, like handling platform-specific dependencies, re-organizing the project structure without losing the contributor's history, testing on multiple platforms, and publishing the library.

11:20 AM: 🎤 Muhammad Salman Bediya - Crucial Performance Issue in Flutter Apps: Memory Leaks

Memory leaks can be hard to spot but have a big impact on the performance of Flutter apps, especially those running for long periods. In this talk, we’ll explore the most common reasons memory leaks happen in Flutter and Dart, focusing on how asynchronous programming and Streams can make them more challenging. You’ll learn practical tips to identify and fix these issues, helping your apps run smoother and more efficiently.

11:40 AM: 🎤 Andrii Raikov - Maximizing Scalability with Go and Redis: A Telemetry Processing Journey

At Delivery Hero, we process 10,000 requests per second using Go and Redis. Join us to learn how this powerful duo handles high-load telemetry data efficiently and cost-effectively, with scalability, resource optimization, and continuous innovation through customized data flows.

12:30 PM: 🎤 Tomek Porozynski - Can You Outsmart an AI? Adventures in Prompt Hacking

In this talk combined with hands-on elements, participants will engage in a series of live prompt hacking challenges, accessible directly through their mobile devices. The workshop begins with simple prompt injection techniques and progressively moves to more sophisticated manipulation strategies. After each successful hack, I'll analyze what made it work and transform these insights into practical defense mechanisms.

Attendees will learn: Common vulnerabilities in AI prompt design, Practical techniques for prompt injection attacks, Essential strategies for securing chatbot applications, Best practices for implementing defensive layers, Real-world examples of prompt security failures and successes

Perfect for developers working with AI models, security enthusiasts, or anyone interested in building safer AI applications. No specialized tools needed - just bring your phone and creativity! You'll leave with concrete techniques for both testing and securing your AI systems against prompt manipulation attacks.

1:10 PM: Lunch 🍔🥤

2:40 PM: 🎤 Cesar Martinez - Domain Driven Design Fundamentals for Frontend Developers

What can we learn from Domain Driven Design and how to start applying its teachings in your frontend codebase.

3:30 PM: 🎤 Vadym Pinchuk - Effortless optimization of Flutter apps: performance tips for developers

In this session, we’ll dive into effortless yet impactful ways to optimize your Flutter applications. Performance improvements don’t always require a full rewrite—sometimes, small adjustments can lead to big gains. We'll explore practical tips and tricks for enhancing app speed, responsiveness, and efficiency with minimal effort. From reducing widget rebuilds to handling large data efficiently and managing state effectively, this talk will provide developers with actionable insights to deliver a smoother user experience. Whether you’re a beginner or an experienced Flutter dev, you’ll walk away with easy-to-apply techniques to optimize your apps without breaking a sweat.

4:20 PM: 🎤 Ian Ballantyne - Generative AI on Mobile and Web with Google AI Edge

Generative AI is no longer limited to execution in the cloud. Small language models, such as Gemma 2B, are quickly becoming small and powerful enough for on-device AI, offering benefits like low latency, offline functionality, privacy, and cost-effectiveness. Google AI Edge, with MediaPipe and LiteRT (formerly Tensorflow Lite), enables the development and deployment of efficient on-device AI models. These frameworks handle the complexities of model execution and hardware acceleration, allowing developers to focus on creating innovative AI experiences.

Think generative AI is just about chatbots? Think again. This talk will go beyond basic conversations with language models and explore how on-device generative AI can be integrated into everyday apps ready to help with tasks, answer questions, and provide creative inspiration, all powered by the information located on-device. Imagine truly useful apps that are quick to respond and still work without an internet connection.

5:10 PM: 🎤 Bogdan Plieshka - Automated Testing Layers in a multidimensional Monorepo: Fast-tracking Quality for hundreds apps

In this talk, I’ll dive into the testing layers that make up our quality pipeline at Zattoo, including static analysis, unit, system, and end-to-end testing.

We’ll discuss the concept of quality gates, shift-left approach, and affected domain recognition, which helps us maintain reliability across a large, dynamic codebase, bringing total quality feedback for contributors to 3 minutes.

I’ll share practices for achieving scalable, fast testing in a high-complexity environment, offering insights for anyone working with large-scale applications or monorepos and looking to streamline QA processes.

Day 3

9:00 AM: Registration & Coffee 🥐 ☕️

10:00 AM: 🎤 Inès Mir & Doruk Deniz Kutukculer - Fellowship of Product. How your team setup affects your experience

Did you know there are 2 types of team formation in tech? These formations can change your experience in the team drastically and you better recognise them early to adjust your expectations from the job. And even more importantly, you need to show different qualities on job interviews to get this job in a particular team formation!

Deniz Doruk Kuetuekcueler, a head of engineering, and Inès Mir, a principal product designer, are trying to figure out how design and engineering can effectively work together in these setups.

10:50 AM: 🎤 Alireza Rahmaty - How we automate the App Release Monitoring at GetYourGuide

App release monitoring (ARM) represents a suite of innovative tools designed to monitor the health and stability of iOS and Android app releases. These tools provide real-time updates by sending notifications to Slack channels and logging the app's status throughout the release process. At GetYourGuide, we have developed an ARM to monitor the rollout of our Android and iOS apps from the moment they are submitted to the App Store & Google Play until they are fully released. We ship releases faster and with more confidence using ARM!

11:40 AM: 🎤 Aleksandr Gorbunov - Flutter for frontenders or There and Back Again

Every developer, regardless of specialization, may encounter the need to create a UI for a client application. The choice of technology may depend on the developer, or it may be pre-determined by the client, as happened in my case.

The peculiarity is that, coming from frontend development in JavaScript, I started building user interfaces in Flutter.

Today, there is a vast number of technologies that enable the development of cross-platform applications. These technologies are evolving rapidly, attracting large communities, and more frequently, companies are adopting them. For example, Flutter is a powerful framework that allows developers to create cross-platform applications.

With a high probability, every developer may encounter the need to use such development tools, and it’s great that frameworks like Flutter come with detailed documentation and extensive community support, making it relatively easy to start developing with them. Although, at first glance, everything might not seem smooth, and the desire to revert to familiar methods may arise.

12:05 PM: 🎤 Muhammad Salman Bediya - Crucial Performance Issue in Flutter Apps: Memory Leaks

Memory leaks can be hard to spot but have a big impact on the performance of Flutter apps, especially those running for long periods. In this talk, we’ll explore the most common reasons memory leaks happen in Flutter and Dart, focusing on how asynchronous programming and Streams can make them more challenging. You’ll learn practical tips to identify and fix these issues, helping your apps run smoother and more efficiently.

12:30 PM: 🎤 Ole Bulbuk - Native GUIs For All

Traditionally native GUIs are highly platform dependent and often specific for one programming language. In this talk we will explore a way to create GUI applications that supports virtually all platforms and any programming language. It is very effective and easy to use, too.

1:10 PM: Lunch 🍔🥤

2:40 PM: 🎤 Nicole Terc - Tap it! Shake it! Fling it! Sheep it! - The Gesture Animations Dance!

Let's have fun with animations, gestures and sensors!

Using Compose Multiplatform, we'll go over how to create animations using gestures and sensor events for Android & iOS. We'll cover some basics like how to get the device motion and position information, how to track gestures in the screen, and how you can combine them with animations to have fun!

After this talk, you'll have a better understanding on how to use the sensor frameworks, how to make your own gesture effects, and how to create interesting animations in an easy way.

Keep it fun, keep it animated!

3:30 PM: 🎤 Andrii Khrystian - From waves to widgets: Sound processing in Flutter

In this talk, we'll explore how to work with sound in Flutter apps. We'll go over the basics of adding sound effects and processing audio to make your apps more interesting. You'll learn how to handle audio files and integrate them smoothly with your Flutter projects. This session is great for anyone looking to add audio features to their apps simply and effectively.

4:20 PM: 🎤 Randy Nel Gupta - From Practice: Migration of an Order Processing System to the Cloud

A case study on how an order processing system, processing 50,000 orders daily for an international retailer spread across multiple continents and jurisdictions, is migrated to the cloud. The legacy system is implemented in PL/SQL and must be migrated during ongoing operations.

The presentation will cover all aspects from testing, monitoring, to development and the application of Site Reliability Engineering.

Furthermore, less technical topics will be introduced, such as the systematic composition of teams to ensure the necessary technical as well as domain-specific expertise.

4:50 PM: 🎤 Wietse Venema - Running open large language models in production with serverless GPUs

Many developers are interested in running open large language models, such as Google's Gemma and Llama. Open models give you full control over the deployment options, the timing of model upgrades, the private data that goes into the model, and the ability to fine-tune on specific tasks such as data extraction. Hugging Face TGI is a popular open-source LLM inference server, and Hugging Face TRL is excellent for fine-tuning. You’ll learn how to build and deploy an application that uses an open model on Google Cloud Run with cost-effective GPUs that scale down to zero instances.

Day 4

9:00 AM: Registration & Coffee 🥐 ☕️

10:00 AM: 🎤 Daniel Stamer & Diana Nanova - Workshop: From Prototype to Production

In this hands-on technical workshop participants will work on a hilarious web service prototype and deploy it to the cloud, set up build and deployment pipelines, extend the code base to leverage GenAI functionality, use SRE practices to effectively operate the application and finally strengthen the security posture of the overall software delivery process to guard against supply chain attacks.

1:10 PM: Lunch 🍔🥤

2:40 PM: 🎤 John Nguyen - Building a Chrome Extension using Gemini and Langchain

In this workshop, you will learn the basics of creating a Google Chrome Extension (which will also work on any Chromium-based Browser). We will build a simple Page summarizer using Bun, Typescript, Gemini, and LangChain. We will learn the anatomy of the manifest.json for building a Chrome Extension, Bun's bundler, how to interact with Gemini, and why LangChain is a good idea here.

3:45 PM: 🎤 Guillaume Vernade - How to make the most of Gemini multimodal capabilities?

We all know that in Tech there are always dozens of way of doing anything. But what if we could only use LLM for a first investigation? Let me show you how I'm trying to solve the mystery of who killed my pond's fishes using the power of Gemini.

Day 5

9:00 AM: Registration & Coffee 🥐 ☕️

10:00 AM: 🎤 Mario Bodemann & Joost van Dijk - Workshop: Passkeys on Android: How to get rid of passwords

Passwords. Or two factors? What about multiple factors? Which email did you register with? Why is 'password123' not working on this side, that is password is shared everywhere else?

If you recognize some of those questions, I am happy to add another couple: What are passkeys? Or how about: How to use passkeys to replace passwords in an Android app?

In this workshop I will walk through the later two questions: How to build an Android App that registers and signs users in, using passkeys. Expect a quick explanation of this fancy new technology, why it will replace passwords and how you can store them either on your mobile devices or on dedicated hardware. Following that, a fictive application and service will be built to show you how to use those passkeys and which moving pieces you will need.

Expect to use you Android Studio with Kotlin and common best practices to build an Android app, talking to the public available backend.

11:05 AM: 🎤 Anton Borries - Workshop: Adding Homescreen Widgets to Flutter Apps

HomeScreen Widgets are a great way to provide more Information to your Users right on their HomeScreens providing more ways for your App to appear in User's lives and help them achieve their goals.

In this Workshop we'll look at the necessary steps needed in order to add HomeScreen Widgets to Flutter Apps using the home_widget package

12:10 PM: 🎤 Elena Grahovac - Workshop: Mastering Multiple Engineering Leadership Roles for Maximum Impact

As an engineering manager or technical leader, navigating multiple roles that demand a diverse set of skills is a common yet challenging part of the job.

In this workshop, we will explore how to effectively balance these multiple roles and responsibilities in a complex engineering environment. Participants will be guided through the creation of their own leadership framework, tailored to adapt to the unique situations and styles of each individual. Beginning with identifying core values and responsibilities, the framework is elaborated into an actionable plan to succeed.

This workshop not only offers an opportunity for reflection on personal and professional development but also provides tools and insights to enhance management capabilities and team dynamics. Join us to cultivate a comprehensive approach to leadership that aligns with your unique role, responsibilities, and personal style.

1:10 PM: Lunch 🍔🥤

2:40 PM: 🎤 Gus Martins - Workshop: Gemma for Everyone: Your First Steps with Open Models and AI

Dive into the world of open models and AI with Gemma! This workshop will guide you through the basics of using Gemma, Google's powerful family of language models. Learn how to harness Gemma's capabilities for tasks like text generation, question answering, and more. We'll also explore how to fine-tune Gemma on your own data, allowing you to create custom AI solutions tailored to your needs. No prior experience with large language models is required!

3:45 PM: 🎤 Shahriyar Rzayev - Learn Flask the hard way: Introduce Architecture Patterns

Flask is a popular and flexible web framework for Python, but building scalable and maintainable Flask applications can be challenging without a solid understanding of architecture patterns. This workshop aims to provide participants with a detailed explanation of applying architecture patterns to Flask projects. By exploring various design principles and best practices, attendees will learn how to structure their Flask applications for improved scalability, modularity, and maintainability.

Focusing on the Repository, Unit of Work, and Use Cases patterns, attendees will gain experience in applying these patterns to enhance code organization, maintainability, and testability. All these layers are wired together using Dependency Injection, which is yet another powerful tool to use in your applications.

The application we are going to build is stored in: https://github.com/ShahriyarR/hexagonal-flask-blog-tutorial

We are going to completely rewrite the official Blog application described in Flask documentation by applying architecture patterns.

All abstraction layers are covered by unit and integration tests, which will give the attendees a detailed view of why it is important to structure the application using architecture patterns.

Speakers

Aleksandr Gorbunov - Smart Steel Technologies (Full Stack Developer)

A skilled developer specializing in JavaScript (JS) and TypeScript (TS), with strong expertise in frontend development. Proficient in the Vue ecosystem (Vue2, Vue3, Composition API, Nuxt 3), using Webpack and Vite for project bundling. Experienced in testing with Vitest, Cypress, and Jest. Adept in CSS preprocessors like SASS and Stylus. Additionally, has solid knowledge of Flutter and experie…

Andrey Sitnik - Evil Martians (Lead Engineer)

With more than 20 years in open source, Andrey Sitnik created a few popular CSS tools (PostCSS, Autoprefixer), local-first framework (Logux), and many small libraries with millions of downloads (like Nano ID).

Andrii Khrystian - Dynatrace (Senior Flutter Developer)

GDG Linz organiser. Senior Flutter Developer at Dynatrace. Public speaker and tech writer

Andrii Raikov - Delivery Hero SE (Principal Software Engineer)

Andrii is a Principal Software Engineer at Delivery Hero. He has a total of 15 years of experience with Ruby and has been very passionate about Go for the last 5 years.

Anton Borries - 1KOMMA5° (Software Engineer)

Anton is a Software Engineer working at 1KOMMA5° He loves building great UI and UX using Flutter. Coming from an Android Background the gap between Flutter and native Features has always tickled his interest. This has lead him into improving the experience of developing HomeScreen Widgets for Flutter Apps

Ash Davies

Google Developer Expert for Android, enthusiastic speaker, lead engineer at ImmobilieenScout24, Kotlin aficionado, spends more time travelling than working.

Daniel Stamer - Google (Cloud Customer Engineer)

Daniel is passionate about building modern cloud-native applications on Google's serverless technologies. He works with digital natives out of Germany’s startup capital Berlin and helps to modernize applications or build brand new ones in the cloud.

Danny Preussler - SoundCloud (Android Platform Lead)

Danny is a developer by heart, living in Berlin and leading the Android team at SoundCloud. He worked for companies like Groupon, Viacom, eBay and Alcatel and started his mobile career long before any Android with Java ME and Blackberry applications. Danny writes and talks about mobile development and testing regularly and is a Google Developer Expert for Android and Kotlin.

Elena Grahovac - FerretDB (Director of Engineering)

Elena has been in software engineering since 2007, focusing on backend systems and infrastructure. Having played the roles of both individual contributor and engineering manager, Elena is passionate about combining technical expertise with strong team collaboration. A dedicated advocate of DevOps practices, she aims to enhance workflows and bring teams together. Elena believes in helping peopl…

Gus Martins - Google (Developer Advocate)

Katya Vinnichenko - Google (Program Manager)

Katya is a Program Manager at Google Developer Relations team. Currently she is leading the Google Developer Groups across Europe, the Middle East and Africa.

Marcin Chudy - LeanCode (Senior Flutter Developer)

Marcin is a Senior Flutter Developer at LeanCode, currently playing tech lead role in a big project for the banking sector. Previously worked with backend, web frontend with React, finally settling on mobile and falling in love with Flutter at first sight. After work, he enjoys dancing salsa and bachata and attends metal concerts. Marcin is a Senior Flutter Developer at LeanCode and has …

Marco Gomiero - Airalo (Senior Android Developer | Kotlin GDE)

Marco is an Android engineer, currently working at Airalo. He is a Google Developer Expert for Kotlin, he loves Kotlin and he has experience with native Android and native iOS development, as well as cross-platform development with Flutter and Kotlin Multiplatform. In his spare time, he writes and maintains open-source code, he shares his dev experience by writing on his blog, speaking a…

Mario Bodeman - Yubico (Android Developer Advocate)

Speaker of talks, coder of code, doer of dones.

Muhammad Bediya

Muhammad Salman is a Senior Software Engineer specializing in mobile app development with a focus on building scalable, high-quality applications using Flutter, React Native, Xamarin, and Swift. With experience leading frontend teams on enterprise-level projects that have reached over 1.5 million users, he brings a strong commitment to creating impactful, user-centered solutions. A dedic…

Nicole Terc

Android GDE, Boardgame lover, videogame addict and origami enthusiast, Nicole self taught herself to code and has been fooling around with the Android ecosystem for more than 10 years. She has participated in a diverse variety of projects for several clients around the world, including video streaming, news, social media and public transport applications. Regardless of what the current adventu…

Ole Bulbuk - Ardan Labs

Ole is a backend engineer since the nineties. He has been working for many companies big and small and seen many projects fail or succeed. He loves to be part of the global Go community and working on projects that make the world a better place. In his spare time he is co-organising the Berlin chapter of GDG Golang, develops open source software and enjoys time with his family.

Oleksii Antypov - DmarcDkim.com (Founder & CEO)

Experienced CTO specializing in early-stage startups. Formerly with Rocket Internet and PocketBook, now focused on accelerating global DMARC adoption. Originally from Ukraine, I relocated to Berlin in 2015 to deepen my expertise in building successful startups from the ground up.

Raphaël VO - Ekino (Senior Software Engineer)

I’m Raphael Vo, a passionate Senior Software Engineer with over 10 years of experience, specializing in Angular and frontend development. I love turning complex ideas into delightful user experiences and tackling challenges creatively and enthusiastically. When I'm not coding, you’ll find me diving into the latest tech trends or enjoying epic board game nights with friends. As an aspiring spea…

Vadim Makeev

Frontend developer in love with the Web, browsers, bicycles, and podcasting. He/him, MDN technical writer, Google Developer Expert.

Alex Mir - mobile.de (Frontend Engineer)

Frontend Engineer at car retail platform mobile.de (part of Adevinta / ex-Ebay)

Alireza Rahmaty - GetYourGuide (Android Developer)

I am Alireza, an Android developer with 6+ years of experience building apps. I have experience building server-driven UI apps, complex UI, localisation and testing, and CI/CDI. I sometimes go hiking and play video games.

Cesar Martinez - Meyer Sound (Web Developer)

Web developer with around 10 years of experience and a passion for software architecture. Currently working at Meyer Sound.

Bogdan Plieshka - Zattoo (Principal Engineer)

Engineer with over a decade of Frontend development experience, passionate about automation, accessibility, and scaling complex systems. Working at Zattoo as a Principal Engineer, focusing on delivering frontend solutions across Web, React, and React Native for streaming media content.Organizer of the React Berlin Meetup, actively contributing to the development community.

Diana Nanova - Google (Customer Engineering Manager)

Diana is a Customer Engineering Manager at Google Cloud. Based in the German tech startup capital Berlin, Diana helps digital native customers and startups across various industries to leverage the capabilities of Google Cloud and loves championing for Google culture.

Doruk Deniz Kutukculer - Zalando (Head of Engineering)

IT professional and a leader with over 15 years of experience in the industry. Currently a Head of Engineering at Zalando.

Guillaume Vernade - Google (AI Dev Rel)

I've been a jack-of-all-trades in the Tech industry, starting as a prototyper building apps on Google Glasses and the first Android watches, then became a Product Owner and an Agile coach. I realized my childhood dream of becoming a video game producer then came back to my other passion: AI.

Ian Ballantyne - Google (AI DevRel)

Ian is a Developer Relations Engineer for AI at Google. Currently he works on generative AI, such as Gemini and Gemma. He is passionate about on-device AI, using technologies such as Google AI Edge to deploy artificial intelligence to web and mobile devices. He has been in Developer Relations at Google for 9 years specializing in helping partners and developers unlock the capability of Google …

Inès Mir - Zalando (Principal Product Designer)

A principal product designer at Zalando and a content creator.

John Nguyen - Eon (Backend Developer)

Fullstack developer with a knack for whipping up code recipes using my secret ingredients: a dash of JavaScript, a pinch of Python, and a whole lot of serverless magic John's journey in software development began as a PHP developer, but he later transitioned to front-end development and became passionate about all things related to Javascript. While working as a data DevOps engineer in a…

Joost van Dijk - Yubico (Developer Advocate)

Joost van Dijk is a developer advocate at Yubico. As the inventor of the YubiKey, Yubico makes secure login easy and available for everyone. Joost focuses on securing digital identities and accelerating the adoption of open authentication standards as part of Yubico’s developer program.

Randy Gupta

Randy is a Google Developer Expert for Cloud and also Organizer of the GDG Düsseldorf. With a professional experience of more 25 years in software development he is focused today on building microservices applications on top of Kubernetes.

Shahriyar Rzayev - Nord Security (Senior Software Engineer)

Senior Software Engineer @ Nord Security. Moving forward on Clean Code and Clean Architecture. Previous accomplishments include contributing to open source, providing technical direction, and sharing knowledge about Clean Code and Architectural patterns. An empathetic team player and mentor. Azerbaijan Python Group Leader. Former QA Engineer and Bug Hunter.

Tomek Porożyński - Atos

Vadym Pinchuk - Sky (Mobile Software Engineer)

Vadym, a seasoned software engineer, possesses a wealth of experience in Android application development. He has skillfully transitioned his expertise to cross-platform development, utilizing Flutter. Throughout his career, Vadym has collaborated with a diverse range of companies, from industry giants like Samsung, Volvo, Bosch, and Instagram to smaller start-ups. Leveraging his extensiv…

Wietse Venema - Google (Google Cloud Engineer)

Wietse Venema is an engineer at Google Cloud. He wrote the O’Reilly book on Cloud Run.

Hosts

Seemran Xec - Sawayo (Software Engineer)

A focused developer possessing professional experience of 6+ years in software development for product-based and service-based industries, with businesses acquiring valuable insight and implementing best practices. Collaborated with startups and other businesses as a freelancer/consultant to build, design, and manage the product. I'm passionate about what I do and a lifelong learner.

Louis Tsai - Zalando SE (GDG Organizer)

Alex Mir - mobile.de (Frontend Engineer)

Frontend Engineer at car retail platform mobile.de (part of Adevinta / ex-Ebay)

Jhoon Saravia - Greenmates (Mobile Engineer)

Software consultant and developer, experienced in Android, Flutter and Full-stack. Interested in working on DEI initiatives as a complement to my core work. Particularly interested in technology, gadgetry, the future, the combination of those three and the impact that driving Diversity, Equity and Inclusion has on all of them both in and out of the workplace.Amateur photographer a…

Matthias Geisler - Thermondo (Senior Software Engineer)

True believer in (Kotlin) Multiplatform and working with it for over 4 years now. Builds solutions for Android. Maintainer and developer of KMock. Co-Organizer of KUG Berlin, GDG Android Berlin, Rust Berlin and XTC Berlin.

Emy Jamalian - Atlas Metrics (Software QA Engineer)

Complete your event RSVP here: https://gdg.community.dev/events/details/google-gdg-berlin-presents-devfest-berlin-2024/.

dbt Meetups are networking events open to all folks working with data! Talks predominantly focus on community members’ experience with dbt, however, you’ll catch presentations on broader topics such as analytics engineering, data stacks, DataOps, modeling, testing, and team structures.

📍 Venue Host: Utopicus Habana (P.º de La Habana, 9, 11, 28036 Madrid) 🍕 Catering: Drinks & Pizza at the place of the event 🤝 Organizer: Astrafy is organizing this event, enabled by the community team at dbt Labs

*To attend, please read the Health and Safety Policy and Terms of Participation: ***https://www.getdbt.com/legal/health-and-safety-policy

🗓️Agenda:

• 18:45 - 19:00 \| Welcome
• 19:00 - 19:20 \| Data validation with dbt: Going beyond testing (Alejandro de la Cruz Lopez - Astrafy)
• 19:25 - 19:45 \| How Airflow + dbt Work Together at Okta (Miquel Angel Andreu Febrer - Okta)
• 19:50 - 20:10 \| Leveraging column-level lineage to scale your dbt projects (Benoit Perigaud - dbt)
• 20:15 - 22h00 \| Networking with drinks & pizzas 🍺 🍕

🗣️Presentation #1: dbt can be leveraged for more than just basic testing. We will dive into advanced data validation techniques that ensure data quality beyond conventional testing in dbt. We will use Recce as a new emerging tool that allows validation checks and improved approval requests.

Speaker bio: Alejandro de la Cruz López is an experienced Data Engineer with a strong background in Data Science and Artificial Intelligence. He has led various data projects, optimizing systems and improving infrastructure for several organizations. Alejandro holds multiple professional certifications and has authored articles on data engineering practices. His work focuses on delivering efficient, scalable data solutions in the cloud.

---

🗣️Presentation #2: In this presentation, Miquel Angel will show how Okta dynamically builds all dbt DAGs from upstream to downstream based on tags and the dbt project structure, automates tests inside the dags, and uses the same warehouse configuration for both dbt runs and tests.

Speaker bio: Data Engineer specialized in ETL, BigData processes, and DevOps

🗣️Presentation #3: Today, we have tools to enforce quality checks on projects, at the model level, like dbt_project_evaluator. Those tools are indispensable to allow teams to scale their dbt transformation. But while we've been focusing on rules at the model level. Could we leverage CLL to also define rules at the column level now? The idea of this talk would be to build an open-source tool and present what problems it can solve.

Speaker bio: Staff Analytics Engineer at dbt Labs

➡️ Join the dbt Slack community: https://www.getdbt.com/community/ 🤝For the best Meetup experience, make sure to join the #local-\ channel in dbt Slack (https://slack.getdbt.com/). ---------------------------------- dbt is the standard in data transformation, used by over 40,000 organizations worldwide. Through the application of software engineering best practices like modularity, version control, testing, and documentation, dbt’s analytics engineering workflow helps teams work more efficiently to produce data the entire organization can trust. Learn more: https://www.getdbt.com/

Register for the Meetup

Date, Time and Location

Date and Time Nov 19, 2024 from 5:30 PM to 8:30 PM

Location The Meetup will take place at Fora Space SOHO located at 33 Broadwick St in London.

Understanding Memory in AI Agents and Agentic Systems

Agentic Systems and extensible compound AI systems are revolutionizing LLM applications, positioning themselves as critical tools in modern AI development. These advanced systems go beyond traditional automation, offering capabilities that drive significant productivity and efficiency gains in enterprise and commercial workflows. However, adopting AI Agents and Agentic Systems at scale poses unique challenges, particularly in ensuring consistent performance, reliability, and scalability.

Central to overcoming these challenges is the role of memory. Memory within AI systems is not only essential for retaining operational data but also for enabling adaptive learning, entity profiling, and customized interactions. Different types of memory, such as short-term and long-term memory, play distinct roles in supporting an agent’s functionality. This talk will delve into the architecture of Agentic Systems and examine how various forms of memory—working memory, data stores, profilers, and toolboxes—contribute to creating robust, efficient, and scalable AI solutions. Attendees will gain insight into how memory is leveraged to enable learning from past executions, personalize interactions, and enhance system capabilities in complex AI applications.

About the Speaker

Richmond Alake is an AI/ML Developer Advocate at MongoDB, where he creates high-quality technical learning content for Developers and MongoDB customers building AI applications. In this role, he provides expert guidance on best practices for developing AI solutions that leverage Large Language Models (LLMs) and MongoDB, as well as offering insights on integrations and other critical aspects of AI development.

How to Unlock More Value from Self-Driving Datasets

AV/ADAS is one of the most advanced fields in Visual AI. However, getting your hands on a high quality dataset can be tough, let alone working with them to get a model to production. In this talk, I will show you the leading methods and tools to help visualize as well take these datasets to the next level. I will demonstrate how to clean and curate AV datasets as well as perform state of the art augmentations using diffusion models to create synthetic data that can empower the self driving car models of the future,

About the Speaker

Daniel Gural is a seasoned Machine Learning Engineer at Voxel51 with a strong passion for empowering Data Scientists and ML Engineers to unlock the full potential of their data.

Benchmarking and Optimization of LLMs

In this session, we’ll focus on the critical role benchmarking plays in optimizing the use of large language models. We’ll dive into how to measure performance across different hardware setups, frameworks, and optimizations such as quantization and attention mechanisms.

About the Speaker

Ciera Fowler is the ML Engineering Lead at Ori, an AI native GPU cloud provider, and an MBA Student at London Business School. Ciera’s works on thought leadership pieces for Ori’s blog and speaking engagements focused on benchmarking and analysis of LLMs. She also posts tutorials and presents at tech meetups around London to help others start building their own LLM powered agents and applications.

Register for the Meetup

Date, Time and Location

Date and Time Nov 19, 2024 from 5:30 PM to 8:30 PM

Location The Meetup will take place at Fora Space SOHO located at 33 Broadwick St in London.

Understanding Memory in AI Agents and Agentic Systems

Agentic Systems and extensible compound AI systems are revolutionizing LLM applications, positioning themselves as critical tools in modern AI development. These advanced systems go beyond traditional automation, offering capabilities that drive significant productivity and efficiency gains in enterprise and commercial workflows. However, adopting AI Agents and Agentic Systems at scale poses unique challenges, particularly in ensuring consistent performance, reliability, and scalability.

Central to overcoming these challenges is the role of memory. Memory within AI systems is not only essential for retaining operational data but also for enabling adaptive learning, entity profiling, and customized interactions. Different types of memory, such as short-term and long-term memory, play distinct roles in supporting an agent’s functionality. This talk will delve into the architecture of Agentic Systems and examine how various forms of memory—working memory, data stores, profilers, and toolboxes—contribute to creating robust, efficient, and scalable AI solutions. Attendees will gain insight into how memory is leveraged to enable learning from past executions, personalize interactions, and enhance system capabilities in complex AI applications.

About the Speaker

Richmond Alake is an AI/ML Developer Advocate at MongoDB, where he creates high-quality technical learning content for Developers and MongoDB customers building AI applications. In this role, he provides expert guidance on best practices for developing AI solutions that leverage Large Language Models (LLMs) and MongoDB, as well as offering insights on integrations and other critical aspects of AI development.

How to Unlock More Value from Self-Driving Datasets

AV/ADAS is one of the most advanced fields in Visual AI. However, getting your hands on a high quality dataset can be tough, let alone working with them to get a model to production. In this talk, I will show you the leading methods and tools to help visualize as well take these datasets to the next level. I will demonstrate how to clean and curate AV datasets as well as perform state of the art augmentations using diffusion models to create synthetic data that can empower the self driving car models of the future,

About the Speaker

Daniel Gural is a seasoned Machine Learning Engineer at Voxel51 with a strong passion for empowering Data Scientists and ML Engineers to unlock the full potential of their data.

Benchmarking and Optimization of LLMs

In this session, we’ll focus on the critical role benchmarking plays in optimizing the use of large language models. We’ll dive into how to measure performance across different hardware setups, frameworks, and optimizations such as quantization and attention mechanisms.

About the Speaker

Ciera Fowler is the ML Engineering Lead at Ori, an AI native GPU cloud provider, and an MBA Student at London Business School. Ciera’s works on thought leadership pieces for Ori’s blog and speaking engagements focused on benchmarking and analysis of LLMs. She also posts tutorials and presents at tech meetups around London to help others start building their own LLM powered agents and applications.