Collaborating on code and software is essential to open science—but it’s not always easy. Join this BoF for an interactive discussion on the real-world challenges of open source collaboration. We’ll explore common hurdles like Python packaging, contributing to existing codebases, and emerging issues around LLM-assisted development and AI-generated software contributions.

We’ll kick off with a brief overview of pyOpenSci—an inclusive community of Pythonistas, from novices to experts—working to make it easier to create, find, share, and contribute to reusable code. We’ll then facilitate small-group discussions and use an interactive Mentimeter survey to help you share your experiences and ideas.

Your feedback will directly shape pyOpenSci’s priorities for the coming year, as we build new programs and resources to support your work in the Python scientific ecosystem. Whether you’re just starting out or a seasoned developer, you’ll leave with clear ways to get involved and make an impact on the broader Python ecosystem in service of advancing scientific discovery.

AI assistants are evolving from simple Q&A bots to intelligent, multimodal, multilingual, and agentic systems capable of reasoning, retrieving, and autonomously acting. In this talk, we’ll showcase how to build a voice-enabled, multilingual, multimodal RAG (Retrieval-Augmented Generation) assistant using Gradio, OpenAI’s Whisper, LangChain, LangGraph, and FAISS. Our assistant will not only process voice and text inputs in multiple languages but also intelligently retrieve information from structured and unstructured data. We’ll demonstrate this with a flight search use case—leveraging a flight database for retrieval and, when necessary, autonomously searching external sources using LangGraph. You will gain practical insights into building scalable, adaptive AI assistants that move beyond static chatbots to autonomous agents that interact dynamically with users and the web.

Block-based programming divides inputs into local arrays that are processed concurrently by groups of threads. Users write sequential array-centric code, and the framework handles parallelization, synchronization, and data movement behind the scenes. This approach aligns well with SciPy's array-centric ethos and has roots in older HPC libraries, such as NWChem’s TCE, BLIS, and ATLAS.

In recent years, many block-based Python programming models for GPUs have emerged, like Triton, JAX/Pallas, and Warp, aiming to make parallelism more accessible for scientists and increase portability.

In this talk, we'll present cuTile and Tile IR, a new Pythonic tile-based programming model and compiler recently announced by NVIDIA. We'll explore cuTile examples from a variety of domains, including a new LLAMA3-based reference app and a port of miniWeather. You'll learn the best practices for writing and debugging block-based Python GPU code, gain insight into how such code performs, and learn how it differs from traditional SIMT programming.

By the end of the session, you'll understand how block-based GPU programming enables more intuitive, portable, and efficient development of high-performance, data-parallel Python applications for HPC, data science, and machine learning.

LLMs are powerful, flexible, easy-to-use... and often wrong. This is a dangerous combination, especially for data analysis and scientific research, where correctness and reproducibility are core requirements. Fortunately, it turns out that by carefully applying LLMs to narrower use cases, we can turn them into surprisingly reliable assistants that accelerate and enhance, rather than undermine, scientific work.

This is not just theory—I’ll showcase working examples of seamlessly integrating LLMs into analytic workflows, helping data scientists build interactive, intelligent applications without needing to be web developers. You’ll see firsthand how keeping LLMs focused lets us leverage their "intelligence" in a way that’s practical, rigorous, and reproducible.

Large language models (LLMs) enable powerful data-driven applications, but many projects get stuck in “proof-of-concept purgatory”—where flashy demos fail to translate into reliable, production-ready software. This talk introduces the LLM software development lifecycle (SDLC)—a structured approach to moving beyond early-stage prototypes. Using first principles from software engineering, observability, and iterative evaluation, we’ll cover common pitfalls, techniques for structured output extraction, and methods for improving reliability in real-world data applications. Attendees will leave with concrete strategies for integrating AI into scientific Python workflows—ensuring LLMs generate value beyond the prototype stage.

Training Large Language Models (LLMs) requires processing massive-scale datasets efficiently. Traditional CPU-based data pipelines struggle to keep up with the exponential growth of data, leading to bottlenecks in model training. In this talk, we present NeMo Curator, an accelerated, scalable Python-based framework designed to curate high-quality datasets for LLMs efficiently. Leveraging GPU-accelerated processing with RAPIDS, NeMo Curator provides modular pipelines for synthetic data generation, deduplication, filtering, classification, and PII redaction—improving data quality and training efficiency.

We will showcase real-world examples demonstrating how multi-node, multi-GPU processing scales dataset preparation to 100+ TB of data, achieving up to 7% improvement in LLM downstream tasks. Attendees will gain insights into configurable pipelines that enhance training workflows, with a focus on reproducibility, scalability, and open-source integration within Python's scientific computing ecosystem.

A lot of data scientists use UMAP to help them quickly visualize and explore complex datasets. This could be exploring large unstructured datasets via neural embeddings, or working on LLM explainability by mapping out Sparse Autoencoder features. Making the visualizations good enough, and compelling enough, to present to end users is much harder. However, if done right a good UMAP plot can be a powerful communication tool, or a rich interactive experience that draws users in. Attendees will come away with a sense of what is possible, and an introduction to open source tools that can make it easy.

This book is essential for anyone eager to understand the groundbreaking advancements in generative AI and its transformative effects across industries, making it a valuable resource for both professional growth and creative inspiration. Generative AI: Disruptive Technologies for Innovative Applications delves into the exciting and rapidly evolving world of generative artificial intelligence and its profound impact on various industries and domains. This comprehensive volume brings together leading experts and researchers to explore the cutting-edge advancements, applications, and implications of generative AI technologies. This volume provides an in-depth exploration of generative AI, which encompasses a range of techniques such as generative adversarial networks, recurrent neural networks, and transformer models like GPT-3. It examines how these technologies enable machines to generate content, including text, images, and audio, that closely mimics human creativity and intelligence. Readers will gain valuable insights into the fundamentals of generative AI, innovative applications, ethical and social considerations, interdisciplinary insights, and future directions of this invaluable emerging technology. Generative AI: Disruptive Technologies for Innovative Applications is an indispensable resource for researchers, practitioners, and anyone interested in the transformative potential of generative AI in revolutionizing industries, unleashing creativity, and pushing the boundaries of what’s possible in artificial intelligence. Audience AI researchers, industry professionals, data scientists, machine learning experts, students, policymakers, and entrepreneurs interested in the innovative field of generative AI.

This workshop is designed to equip software engineers with the skills to build and iterate on generative AI-powered applications. Participants will explore key components of the AI software development lifecycle through first principles thinking, including prompt engineering, monitoring, evaluations, and handling non-determinism. The session focuses on using multimodal AI models to build applications, such as querying PDFs, while providing insights into the engineering challenges unique to AI systems. By the end of the workshop, participants will know how to build a PDF-querying app, but all techniques learned will be generalizable for building a variety of generative AI applications.

If you're a data scientist, machine learning practitioner, or AI enthusiast, this workshop can also be valuable for learning about the software engineering aspects of AI applications, such as lifecycle management, iterative development, and monitoring, which are critical for production-level AI systems.

Drawing from practical lessons learned while building and maintaining customer-facing AI applications across Bloomberg Law, Bloomberg Tax, and Bloomberg Government, this talk explores the unique position of data-rich enterprises in today’s rapidly evolving AI landscape. These organizations possess deep reserves of proprietary data that foundational models have not seen during training. This talk will examine the strategic and technical considerations of leveraging such exclusive datasets, and how these enterprises can meaningfully participate in the AI transformation without developing their own models.

Through the use of NetworkX's API, tutorial participants will learn about the basics of graph theory and its use in applied network science. Starting with a computationally-oriented definition of a graph and its associated methods, we will progress through the following concepts: path and structure finding, visualization, and graph storage on disk. We will also offer tutorial participants the option of one advanced topic overview, including the use of graphs alongside LLMs for knowledge retrieval, scalable alternatives to NetworkX including cuGraph, and the use of linear algebraic translation of graph problems to speed up computations.

Todd Olson joins me to talk about making analytics worth paying for and relevant in the age of AI. The CEO of Pendo, an analytics SAAS company, Todd shares how the company evolved to support a wider audience by simplifying dashboards, removing user roadblocks, and leveraging AI to both generate and explain insights. We also talked about the roles of product management at Pendo. Todd views AI product management as a natural evolution for adaptable teams and explains how he thinks about hiring product roles in 2025. Todd also shares how he thinks about successful user adoption of his product around “time to value” and “stickiness” over vanity metrics like time spent. 

Highlights/ Skip to:

How Todd has addressed analytics apathy over the past decade at Pendo (1:17) Getting back to basics and not barraging people with more data and power (4:02) Pendo’s strategy for keeping the product experience simple without abandoning power users (6:44) Whether Todd is considering using an LLM (prompt-based) answer-driven experience with Pendo's UI (8:51) What Pendo looks for when hiring product managers right now, and why (14:58) How Pendo evaluates AI product managers, specifically (19:14) How Todd Olson views AI product management compared to traditional software product management (21:56) Todd’s concerns about the probabilistic nature of AI-generated answers in the product UX (27:51) What KPIs Todd uses to know whether Pendo is doing enough to reach its goals (32:49)   Why being able to tell what answers are best will become more important as choice increases (40:05)

Quotes from Today’s Episode

“Let’s go back to classic Geoffrey Moore Crossing the Chasm, you’re selling to early adopters. And what you’re doing is you’re relying on the early adopters’ skill set and figuring out how to take this data and connect it to business problems. So, in the early days, we didn’t do anything because the market we were selling to was very, very savvy; they’re hungry people, they just like new things. They’re getting data, they’re feeling really, really smart, everything’s working great. As you get bigger and bigger and bigger, you start to try to sell to a bigger TAM, a bigger audience, you start trying to talk to the these early majorities, which are, they’re not early adopters, they’re more technology laggards in some degree, and they don’t understand how to use data to inform their job. They’ve never used data to inform their job. There, we’ve had to do a lot more work.” Todd (2:04 - 2:58) “I think AI is amazing, and I don’t want to say AI is overhyped because AI in general is—yeah, it’s the revolution that we all have to pay attention to. Do I think that the skills necessary to be an AI product manager are so distinct that you need to hire differently? No, I don’t. That’s not what I’m seeing. If you have a really curious product manager who’s going all in, I think you’re going to be okay. Some of the most AI-forward work happening at Pendo is not just product management. Our design team is going crazy. And I think one of the things that we’re seeing is a blend between design and product, that they’re always adjacent and connected; there’s more sort of overlappiness now.” Todd (22:41 - 23:28) “I think about things like stickiness, which may not be an aggregate time, but how often are people coming back and checking in? And if you had this companion or this agent that you just could not live without, and it caused you to come into the product almost every day just to check in, but it’s a fast check-in, like, a five-minute check-in, a ten-minute check-in, that’s pretty darn sticky. That’s a good metric. So, I like stickiness as a metric because it’s measuring [things like], “Are you thinking about this product a lot?” And if you’re thinking about it a lot, and like, you can’t kind of live without it, you’re going to go to it a lot, even if it’s only a few minutes a day. Social media is like that. Thankfully I’m not addicted to TikTok or Instagram or anything like that, but I probably check it nearly every day. That’s a pretty good metric. It gets part of my process of any products that you’re checking every day is pretty darn good. So yeah, but I think we need to reframe the conversation not just total time. Like, how are we measuring outcomes and value, and I think that’s what’s ultimately going to win here.” Todd (39:57)

Links

LinkedIn: https://www.linkedin.com/in/toddaolson/  X: https://x.com/tolson  [email protected] 

Large Language Models (LLMs) have revolutionized natural language processing, but they come with limitations such as hallucinations and outdated knowledge. Retrieval-Augmented Generation (RAG) is a practical approach to mitigating these issues by integrating external knowledge retrieval into the LLM generation process.

This tutorial will introduce the core concepts of RAG, walk through its key components, and provide a hands-on session for building a complete RAG pipeline. We will also cover advanced techniques, such as hybrid search, re-ranking, ensemble retrieval, and benchmarking. By the end of this tutorial, participants will be equipped with both the theoretical understanding and practical skills needed to build robust RAG pipeline.

In this tutorial, you will learn how to integrate Large Language Models (LLMs) directly into Python programs as thoughtfully-designed core components of the program rather than bolt-on additions. This hands-on session teaches design principles and practical techniques for incorporating LLM outputs into program control flow. We will use LlamaBot, an open-source Python interface to LLMs, focusing on local execution with local and efficient models.

Supported by Our Partners •⁠ WorkOS — The modern identity platform for B2B SaaS. •⁠ Statsig ⁠ — ⁠ The unified platform for flags, analytics, experiments, and more. • Sonar —  Code quality and code security for ALL code.  — What happens when a company goes all in on AI? At Shopify, engineers are expected to utilize AI tools, and they’ve been doing so for longer than most. Thanks to early access to models from GitHub Copilot, OpenAI, and Anthropic, the company has had a head start in figuring out what works. In this live episode from LDX3 in London, I spoke with Farhan Thawar, VP of Engineering, about how Shopify is building with AI across the entire stack. We cover the company’s internal LLM proxy, its policy of unlimited token usage, and how interns help push the boundaries of what’s possible. In this episode, we cover: • How Shopify works closely with AI labs • The story behind Shopify’s recent Code Red • How non-engineering teams are using Cursor for vibecoding • Tobi Lütke’s viral memo and Shopify’s expectations around AI • A look inside Shopify’s LLM proxy—used for privacy, token tracking, and more • Why Shopify places no limit on AI token spending  • Why AI-first isn’t about reducing headcount—and why Shopify is hiring 1,000 interns • How Shopify’s engineering department operates and what’s changed since adopting AI tooling • Farhan’s advice for integrating AI into your workflow • And much more! — Timestamps (00:00) Intro (02:07) Shopify’s philosophy: “hire smart people and pair with them on problems” (06:22) How Shopify works with top AI labs  (08:50) The recent Code Red at Shopify (10:47) How Shopify became early users of GitHub Copilot and their pivot to trying multiple tools (12:49) The surprising ways non-engineering teams at Shopify are using Cursor (14:53) Why you have to understand code to submit a PR at Shopify (16:42) AI tools' impact on SaaS  (19:50) Tobi Lütke’s AI memo (21:46) Shopify’s LLM proxy and how they protect their privacy (23:00) How Shopify utilizes MCPs (26:59) Why AI tools aren’t the place to pinch pennies (30:02) Farhan’s projects and favorite AI tools (32:50) Why AI-first isn’t about freezing headcount and the value of hiring interns (36:20) How Shopify’s engineering department operates, including internal tools (40:31) Why Shopify added coding interviews for director-level and above hires (43:40) What has changed since Spotify added AI tooling  (44:40) Farhan’s advice for implementing AI tools — The Pragmatic Engineer deepdives relevant for this episode: • How Shopify built its Live Globe for Black Friday • Inside Shopify's leveling split • Real-world engineering challenges: building Cursor • How Anthropic built Artifacts — See the transcript and other references from the episode at ⁠⁠https://newsletter.pragmaticengineer.com/podcast⁠⁠ — Production and marketing by ⁠⁠⁠⁠⁠⁠⁠⁠https://penname.co/⁠⁠⁠⁠⁠⁠⁠⁠. For inquiries about sponsoring the podcast, email [email protected].

Get full access to The Pragmatic Engineer at newsletter.pragmaticengineer.com/subscribe

Está no ar, o Data Hackers News !! Os assuntos mais quentes da semana, com as principais notícias da área de Dados, IA e Tecnologia, que você também encontra na nossa Newsletter semanal, agora no Podcast do Data Hackers !! Aperte o play e ouça agora, o Data Hackers News dessa semana ! Para saber tudo sobre o que está acontecendo na área de dados, se inscreva na Newsletter semanal: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://www.datahackers.news/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Conheça nossos comentaristas do Data Hackers News: Inscrições do Data Hackers Challenge 2025 Live de Abertura Conheça nossos comentaristas do Data Hackers News: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Monique Femme⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ Demais canais do Data Hackers: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Site⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Linkedin⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Instagram⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Tik Tok⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠You Tube⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠

In this keynote, Peeyush Rai and Vikram Koka will be walking through how Airflow is being used as part of a Agentic AI platform servicing insurance companies, which runs on all the major public clouds, leveraging models from Open AI, Google (Gemini), AWS (Claude and Bedrock). This talk walks through the details of the actual end user business workflow including gathering relevant financial data to make a decision, as well as the tricky challenge of handling AI hallucinations, with new Airflow capabilities such as “Human in the loop”. This talk offers something for both business and technical audiences. Business users will get a clear view of what it takes to bring an AI application into production and how to align their operations and business teams with an AI enabled workflow. Meanwhile, technical users will walk away with practical insights on how to orchestrate complex business processes enabling a seamless collaboration between Airflow, AI Agents and Human in the loop.

Join us to explore the DAG Upgrade Agent. Developed with Google Agent Development Kit and powered by Gemini, the DAG Upgrade Agent uses a rules-based framework to analyze DAG code, identify compatibility issues between core airflow and provider package versions, and generates precise upgrade recommendations and automated code conversions. Perfect for upcoming Airflow 3.0 migrations.

KP Division of Research uses Airflow as a central technology for integrating diverse technologies in an agile setting. We wish to present a set of use-cases for AI/ML workloads, including imaging analysis (tissue segmentation, mammography), NLP (early identification of psychosis), LLM processing (identification of vessel diameter from radiological impressions), and other large data processing tasks. We create these “short-lived” project workflows to accomplish specific aims, and then may never run the job again, so leveraging generalized patterns are crucial to quickly implementing these jobs. Our Advanced Computational Infrastructure is comprised of multiple Kubernetes clusters, and we use Airflow to democratize the use of our batch level resources in those clusters. We use Airflow form-based parameters to deploy pods running R and Python scripts where generalized parameters are injected into scripts that follow internal programming patterns. Finally, we also leverage Airflow to create headless services inside Kubernetes for large computational workloads (Spark & H2O) that subsequent pods consume ephemerally.

One of the exciting new features in Airflow 3 is internationalization (i18n), bringing multilingual support to the UI and making Airflow more accessible to users worldwide. This talk will highlight the UI changes made to support different languages, including locale-aware adjustments. We’ll discuss how translations are contributed and managed — including the use of LLMs to accelerate the process — and why human review remains an essential part of it. We’ll present the i18n policy designed to ensure long-term maintainability, along with the tooling developed to support it. Finally, you’ll learn how to get involved and contribute to Airflow’s global reach by translating or reviewing content in your language.