talk-data.com

Welcome to the PyData Berlin September meetup!

We would like to welcome you all starting from 18:30. There will be food and drinks. The talks begin around 19.00.

Please provide your first and last name for the registration because this is required for the venue's entry policy. If you cannot attend, please cancel your spot so others can join as the space is limited.

Host: SPICED Academy is excited to welcome you for this month's version of PyData. At SPICED Academy, we connect the next tech success stories to their futures. Our intensive bootcamps in Data Science, Data Analytics, UX/UI, Java Development and Web Development teach the most up-to-date, in-demand skills on the market. **************************************************************************

The Lineup for the evening

Talk 1: FAir Interpolation Method (FAIM) for Trade-Offs between Mutually Exclusive Algorithmic Fairness Criteria in Machine Learning and Law? Abstract: Ensuring fairness in AI is crucial but challenging, as different fairness criteria often conflict. This talk introduces the FAir Interpolation Method (FAIM), a novel algorithm using optimal transport to balance competing fairness measures, such as ‘calibration within groups’ and ‘balance for the positive/negative class’. FAIM aims to adjust unfair predictions to meet a weighted combination of fairness standards, and its effectiveness is demonstrated with synthetic credit score data, COMPAS data, and e-commerce recommendations. FAIM could help align AI systems with legal standards, including recent EU regulations. Speaker: Alex Loosley is a Responsible AI Scientist and Engineer with experience in developing algorithms for crop protection, document intelligence, and fashion fit recommendations. Outside of his AI work, he referees professional ice hockey in Germany.

Talk 2: AI on a Microbudget - Methods of Machine Learning Miniaturization? Abstract: Current progress in AI has seen remarkable capabilities emerging from simple prediction tasks – if we scale them massively. Surprisingly, we get sparks of reasoning and intelligence in a model that was trained to do little more than masked word prediction. Since that realization the AI field has pursued ever larger models, trained at “eye-watering” cost. If scaling is all you need – does it follow that, in practice, money is all you need? In this talk we explore ideas for the rest of us, the GPU-poor. Taking examples from language processing and computer vision, we’ll show you how to make do with less – less computing power, less person power, less data – while still building powerful models. We will introduce a set of methods and open source tools for the efficient reuse and miniaturization of models, including transfer learning and fine-tuning, knowledge distillation, and model quantization. Our talk aims to provide an overview for ML practitioners, draws from our combined project experience, and is accompanied by a repository of code examples to get you started with building AI on a microbudget. Speaker: Katharina Rasch is a data scientist and computer vision engineer with a PhD in Computer Science from KTH Stockholm. She currently freelances in Berlin and also works as a teacher. Christian Staudt is a data scientist with 8 years of freelance experience. He focuses on machine learning from prototype to deployment, contributes to open source and has organized PyData community events.

Lightning talks There will be slots for 2-3 Lightning Talks (3-5 Minutes for each). Kindly let us know if you would like to present something at the start of the meetup :)

*** NumFOCUS Code of Conduct THE SHORT VERSION Be kind to others. Do not insult or put down others. Behave professionally. Remember that harassment and sexist, racist, or exclusionary jokes are not appropriate for NumFOCUS. All communication should be appropriate for a professional audience including people of many different backgrounds. Sexual language and imagery are not appropriate. NumFOCUS is dedicated to providing a harassment-free community for everyone, regardless of gender, sexual orientation, gender identity, and expression, disability, physical appearance, body size, race, or religion. We do not tolerate harassment of community members in any form. Thank you for helping make this a welcoming, friendly community for all. If you haven't yet, please read the detailed version here: https://numfocus.org/code-of-conduct

PLEASE REGISTER HERE: https://meetup-september10.kickstartai-events.org

PyData Eindhoven & Kickstart.ai We are very happy to announce that we are organizing the next PyData Eindhoven meetup in collaboration with Kickstart.ai! We will host this meetup at the AI Innovation Center, located at the High Tech Campus in Eindhoven.

This meetup is set to take place on September 10, 2024, at the Al Innovation Center in the High Tech Campus, Eindhoven. Mark your calendar and join us for an evening of insights, networking, and cutting-edge technology! Learn the ins and outs of cutting-edge Al and machine learning with our expert speakers, Merel Theisen (Principal Software Engineer at QuantumBlack) and Emilio Oldenziel (Machine Learning Engineer at Eraneos). Discover how software engineering principles can elevate machine learning projects and how Al is optimizing rail traffic control through innovative solutions. Come and connect with fellow Al experts at our Cutting-Edge Al and Machine Learning Innovations meetup, where you'll have the chance to learn from top experts in the field and share your own experiences. Our events provide a platform for collaboration and knowledge sharing, helping us all to advance in the fast-evolving world of Al. And after an evening of learning, enjoy networking drinks and bites, where our speakers will be available to chat and answer any questions you have about Al and machine learning.

Program

• 17:30 – 18:00 Doors Open
• 18:00 – 18:10 🎤 Welcome
• 18:10 – 19:00 🎤 Merel Theisen Principal Software Engineer, QuantumBlack - Embedding Software Engineering Best Practices into Machine Learning Projects with Kedro
• 19:00 – 19:15 🍕 Food
• 19:15 – 20:00 🎤 Emilio Oldenziel Machine Learning Engineer, Eraneos - Optimizing Rail Traffic Control using a Digital Twin and Reinforcement Learning
• 20:00-21:00 🥤 Drinks
• ---

Embedding Software Engineering Best Practices into Machine Learning Projects with Kedro In this talk, I will explore how software engineering best practices such as modularity, separation of concerns, testability, and reproducibility can elevate the quality and deployability of machine learning projects. Focusing on the Kedro framework, I’ll uncover how these principles integrate into data workflows, making complex projects more manageable and scalable. Attendees will gain practical insights into improving project design, ensuring code quality, and facilitating smoother transitions to production environments. No extensive software engineering background is required, making this an accessible and informative session for all data professionals looking to enhance their knowledge of software principles through Kedro. About I am a Principal Software Engineer at QuantumBlack, where I am currently the tech lead of Kedro, an open-source project part of the Linux Foundation. I have over eight years of experience in the software industry, with most of my career focused on backend product engineering. I am passionate about building products that solve real user problems, and I care deeply about creating robust, well-tested software that follows good engineering principles. I am also a strong advocate for open-source software, and I find working with the community to be both inspiring and energising.

---- Optimizing Rail Traffic Control using a Digital Twin and Reinforcement Learning In this talk, I will do a deep dive into one of our recent customer cases. In the customer’s railway network, train frequency is high, with trains departing every 3 minutes at some stations. Consequently, even a small disruption can affect the punctuality of many subsequent trains. Train dispatchers are tasked with resolving conflicts efficiently to minimize delays. I will discuss how a Digital Twin and using Reinforcement Learning (RL) can support dispatchers in making smarter decisions and the challenges of implementing RL solutions. I will demonstrate how combining scientific and practical RL knowledge reduced delays by over 58,000 minutes annually. About Emilio is an expert in the field of machine learning. At the Eraneos Data & AI practice he is responsible for advising on, developing and implementing AI solutions. He has a background in computing science and has worked on AI projects for companies like Porsche, Deutsche Bahn, Enexis and HTM. Transport and Logistics is one of his industry focuses, where he sees a lot of value in applying AI.

This is a place for an in-person meeting in Chilworth. Check out the other one for a virtual meetup on MS Teams.

!!!!! ATTENTION Since September, we've been having meetups at Spectrum IT's office - Gamma House, Enterprise Rd, Chilworth, Southampton SO16 7NS https://maps.app.goo.gl/66tL8HfgFnoiQEh66

Agenda

18:00: Welcome and Intros Tori Tompkins - Beginner's Guide to Large Language Models (LLMs)

pizza/networking Raffle

Tori Tompkins - Beginner's Guide to Large Language Models (LLMs) You've probably noticed a big change in the data world lately — everybody's buzzing about Generative AI and Large Language Models (LLMs). This session will explore everything you need to know about LLMs and Generative AI. Starting with the basics, we will cover how they actually work and how far they have come. Delving deeper, we'll discuss the capabilities and limitations of LLMs, providing insight into what they can and cannot achieve as well as address the ethical considerations and potential risks associated with their usage. We will end with use cases, a demo and what it takes to actually productise a LLM workflow.

Tori Tompkins Senior Data Science Consultant at Advancing Analytics Tori is a Senior Data Science Consultant at Advancing Analytics who has worked on a variety of machine learning and data science projects but mostly with Azure technologies, Databricks, MLFlow, Scikit-learn, PyTorch and NetworkX. She is a regular presenter of YouTube series Data Science Moments and founder of Girls Code Too UK.

*** A massive thanks to our sponsor Spectrum IT. Without them, we could not host this event for free. Spectrum IT kindly provided funding for the room, pizza and a £20 Amazon Voucher for in-person attendees. ***

The 3th edition of the Belgium dbt Meetup will take place in Brussels on September 28th.

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, data ops, modeling, testing, and team structures.

🏠Venue Host: Astrafy, Silversquare Delta, Av. Arnaud Fraiteur 15, Ixelles 🤝Organizers: Charles Verleyen & Sam Debruyn

To attend, please read the Required Participation Language for In-Person Events with dbt Labs: https://bit.ly/3QIJXFb

Our venue has capacity limits, so please only RSVP if you intend to come and reach out send a message in #local-belgium on Slack if you need to cancel last minute or change your RSVP status on the Meetup to "Not Going."

📝Agenda:

• 17h45: welcome with food & drinks
• 18h30: Why dbt will one day be bigger than Spark - Kris Peeters, CEO at Data Minded
• 19h15: Bring your own compute: benefits & pitfalls of developing analytics with DuckDB - Frederik Druyts, Consultant at Datashift
• 19h45: break
• 20h: Speed up ML development with python models in dbt - Dorian Van den Heede, Machine Learning Engineer at Dataroots
• 20h45: networking & drinks

🗣Presentation #1: Why dbt will one day be bigger than Spark - Kris Peeters\, CEO at Data Minded In this talk, I will discuss how the data world is moving from distributed processing frameworks like Spark to more high-level tooling like dbt. It is based on a blog I wrote more than 2 years ago, and we see the transition happen today: https://medium.com/datamindedbe/why-dbt-will-one-day-be-bigger-than-spark-2225cadbdad0

Kris is the founder and CEO of Data Minded. We serve large enterprises and growing startups by designing, building and running data initiatives in the cloud.

🗣Presentation #2: Bring your own compute: benefits & pitfalls of developing analytics with DuckDB - Frederik Druyts\, Consultant at Datashift When was the last time you really needed that 16-machine cluster to quickly check an assumption about your data while writing a new model? Developers often have powerful machines sitting right on their desks, though they rarely use any of that power when doing analytics engineering.

In this session, we will explore how we can leverage DuckDB to efficiently fetch data from cloud storage and build, explore, and test a data model all from a local environment. We will discuss the benefits and pitfalls of this approach, including how it can help us reduce the cost and complexity of our analytics engineering efforts, as well as its potential limitations.

Frederik is skilled at designing, implementing and maintaining end-to-end data solutions. As an analytical mind with a heart for technology, he wants to help the right people ask the right questions and get the right answers from data they can trust. He achieves impact on the intersection of grasping business needs, mastering toolsets both time-tested and new, and writing clean SQL.

🗣Presentation #3: Speed up ML development with python models in dbt - Dorian Van den Heede, Machine Learning Engineer at Dataroots In this talk, Dorian will share his experience on do's and don'ts of python models within dbt for ML and demonstrate a few design patterns to speed up development responsibly within dbt. While SQL is still the preferred language within dbt, python models bring more expressive power to your transformation pipelines. Exactly what we want and need as machine learning engineers for our data preparation! You can go even further to defining your entire ML production pipeline within dbt, inference included.

Dorian (28y) is a Machine Learning Engineer at Dataroots. With his background in Computer Science Engineering and 5 years of experience at multiple clients, bringing machine learning models to production according to software best practices is what he thrives at. Aside of that, he was involved in the AI music band Beatroots, writes punk rock songs on his guitar and can be found blitzing chess moves online.

➡️ Join the dbt Slack community: https://www.getdbt.com/community/ 🤝 For the best Meetup experience, make sure to join the #local-belgium channel in dbt Slack (https://slack.getdbt.com/)!

dbt is a data transformation framework that lets analysts and engineers collaborate using their shared knowledge of SQL. 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/

Zoom Link

https://voxel51.com/computer-vision-events/ai-ml-data-science-meetup-sept-7/

Monitoring Large Language Models (LLMs) in Production

Just like with all machine learning models, once you put an LLM in production you’ll probably want to keep an eye on how it’s performing. Observing key language metrics about user interaction and responses can help you craft better prompt templates and guardrails for your applications. This talk will take a look at what you might want to be looking at once you deploy your LLMs.

Sage Elliott is a Technical Evangelist – Machine Learning & MLOps at WhyLabs. He enjoys breaking down the barrier to AI observability and talking to amazing people in the AI community.

Neural Residual Radiance Fields for Streamably Free-Viewpoint Videos

The success of the Neural Radiance Fields (NeRFs) for modeling and free-view rendering static objects has inspired numerous attempts on dynamic scenes.Current techniques that utilize neural rendering for facilitating freeview videos (FVVs) are restricted to either offline rendering or are capable of processing only brief sequences with minimal motion. In this paper, we present a novel technique, Residual Radiance Field or ReRF, as a highly compact neural representation to achieve real-time FVV rendering on long-duration dynamic scenes.

Minye Wu – Postdoctoral researcher, KU Leuven

Egoschmema: A Dataset for Truly Long-Form Video Understanding

Introducing EgoSchema, a very long-form video question-answering dataset, and benchmark to evaluate long video understanding capabilities of modern vision and language systems. Derived from Ego4D, EgoSchema consists of over 5000 human curated multiple choice question answer pairs, spanning over 250 hours of real video data, covering a very broad range of natural human activity and behavior.

Karttikeya is a PhD student in Computer Science at the Department of Electrical Engineering & Computer Sciences (EECS) at University of California, Berkeley advised by Prof. Jitendra Malik. Earlier, he held a visiting researcher position at Meta AI where he collaborated with Dr. Christoph Feichtenhofer and team.

Zoom Link

https://voxel51.com/computer-vision-events/ai-ml-data-science-meetup-sept-7/

Monitoring Large Language Models (LLMs) in Production Just like with all machine learning models, once you put an LLM in production you’ll probably want to keep an eye on how it’s performing. Observing key language metrics about user interaction and responses can help you craft better prompt templates and guardrails for your applications. This talk will take a look at what you might want to be looking at once you deploy your LLMs.

Sage Elliott is a Technical Evangelist – Machine Learning & MLOps at WhyLabs. He enjoys breaking down the barrier to AI observability and talking to amazing people in the AI community.

Neural Residual Radiance Fields for Streamably Free-Viewpoint Videos

The success of the Neural Radiance Fields (NeRFs) for modeling and free-view rendering static objects has inspired numerous attempts on dynamic scenes.Current techniques that utilize neural rendering for facilitating freeview videos (FVVs) are restricted to either offline rendering or are capable of processing only brief sequences with minimal motion. In this paper, we present a novel technique, Residual Radiance Field or ReRF, as a highly compact neural representation to achieve real-time FVV rendering on long-duration dynamic scenes.

Minye Wu – Postdoctoral researcher, KU Leuven

Egoschmema: A Dataset for Truly Long-Form Video Understanding

Introducing EgoSchema, a very long-form video question-answering dataset, and benchmark to evaluate long video understanding capabilities of modern vision and language systems. Derived from Ego4D, EgoSchema consists of over 5000 human curated multiple choice question answer pairs, spanning over 250 hours of real video data, covering a very broad range of natural human activity and behavior.

Karttikeya is a PhD student in Computer Science at the Department of Electrical Engineering & Computer Sciences (EECS) at University of California, Berkeley advised by Prof. Jitendra Malik. Earlier, he held a visiting researcher position at Meta AI where he collaborated with Dr. Christoph Feichtenhofer and team.

Zoom Link

https://voxel51.com/computer-vision-events/ai-ml-data-science-meetup-sept-7/

Monitoring Large Language Models (LLMs) in Production

Just like with all machine learning models, once you put an LLM in production you’ll probably want to keep an eye on how it’s performing. Observing key language metrics about user interaction and responses can help you craft better prompt templates and guardrails for your applications. This talk will take a look at what you might want to be looking at once you deploy your LLMs.

Sage Elliott is a Technical Evangelist – Machine Learning & MLOps at WhyLabs. He enjoys breaking down the barrier to AI observability and talking to amazing people in the AI community.

Neural Residual Radiance Fields for Streamably Free-Viewpoint Videos

The success of the Neural Radiance Fields (NeRFs) for modeling and free-view rendering static objects has inspired numerous attempts on dynamic scenes.Current techniques that utilize neural rendering for facilitating freeview videos (FVVs) are restricted to either offline rendering or are capable of processing only brief sequences with minimal motion. In this paper, we present a novel technique, Residual Radiance Field or ReRF, as a highly compact neural representation to achieve real-time FVV rendering on long-duration dynamic scenes.

Minye Wu – Postdoctoral researcher, KU Leuven

Egoschmema: A Dataset for Truly Long-Form Video Understanding

Introducing EgoSchema, a very long-form video question-answering dataset, and benchmark to evaluate long video understanding capabilities of modern vision and language systems. Derived from Ego4D, EgoSchema consists of over 5000 human curated multiple choice question answer pairs, spanning over 250 hours of real video data, covering a very broad range of natural human activity and behavior.

Karttikeya is a PhD student in Computer Science at the Department of Electrical Engineering & Computer Sciences (EECS) at University of California, Berkeley advised by Prof. Jitendra Malik. Earlier, he held a visiting researcher position at Meta AI where he collaborated with Dr. Christoph Feichtenhofer and team.

Zoom Link

https://voxel51.com/computer-vision-events/ai-ml-data-science-meetup-sept-7/

Monitoring Large Language Models (LLMs) in Production Just like with all machine learning models, once you put an LLM in production you’ll probably want to keep an eye on how it’s performing. Observing key language metrics about user interaction and responses can help you craft better prompt templates and guardrails for your applications. This talk will take a look at what you might want to be looking at once you deploy your LLMs.

Sage Elliott is a Technical Evangelist – Machine Learning & MLOps at WhyLabs. He enjoys breaking down the barrier to AI observability and talking to amazing people in the AI community.

Neural Residual Radiance Fields for Streamably Free-Viewpoint Videos

The success of the Neural Radiance Fields (NeRFs) for modeling and free-view rendering static objects has inspired numerous attempts on dynamic scenes.Current techniques that utilize neural rendering for facilitating freeview videos (FVVs) are restricted to either offline rendering or are capable of processing only brief sequences with minimal motion. In this paper, we present a novel technique, Residual Radiance Field or ReRF, as a highly compact neural representation to achieve real-time FVV rendering on long-duration dynamic scenes.

Minye Wu – Postdoctoral researcher, KU Leuven

Egoschmema: A Dataset for Truly Long-Form Video Understanding

Introducing EgoSchema, a very long-form video question-answering dataset, and benchmark to evaluate long video understanding capabilities of modern vision and language systems. Derived from Ego4D, EgoSchema consists of over 5000 human curated multiple choice question answer pairs, spanning over 250 hours of real video data, covering a very broad range of natural human activity and behavior.

Karttikeya is a PhD student in Computer Science at the Department of Electrical Engineering & Computer Sciences (EECS) at University of California, Berkeley advised by Prof. Jitendra Malik. Earlier, he held a visiting researcher position at Meta AI where he collaborated with Dr. Christoph Feichtenhofer and team.