Open AI + Data Forum @ Open Source Summit Europe 2023
2023-09-19..2023-09-21
Bilbao, Spain
[Open AI + Data Forum @ Open Source Summit North America 2023, Open AI + Data Forum @ Open Source Summit North America 2024, Open AI + Data Forum @ Open Source Summit North America 2022, Open Source Summit Europe 2022: Open AI + Data Forum, AI_dev: Open Source GenAI & ML Summit Europe 2024, Open Source Summit Latin America 2022: Open AI + Data Forum]
who it's for
ml platform engineers, machine learning engineers, data engineers, data scientists, site reliability engineers, platform engineers
topics
serverless federated learning; local differential privacy; python pickle vulnerabilities; model registry and notebook diffs; scaling feature stores; time-series one-table-per-sensor; Kaskada time-travel features; delta-rs python bindings; deploying open-source LLMs with Ray Serve; vector search with Qdrant; explainable AI with SHAP and LIME; distributed I/O for LLM training
| title | author | length |
|---|---|---|
| Understanding the Legal Context of AI | Van Lindberg | 42m |
Recent lawsuits name GitHub Copilot, Stable Diffusion, DALL‑E and Midjourney and challenge training on copyrighted material while courts treat training, model… | ||
| FedLess: Secure and Scalable Serverless Federated Learning | Mohak Chadha | 31m |
FedLess is an open‑source, cloud‑agnostic FL framework that runs client code as serverless functions on AWS Lambda, Google Cloud Functions, Azure Functions and… | ||
| Poisoned Pickles Make You Ill | Adrian Gonzalez-Martin | 32m |
Claim: Python pickles can carry arbitrary code so a serialized scikit learn model can be poisoned pickle to exfiltrate secrets. Demo modifies the model's… | ||
| MLOps: Why DevOps Solutions Fall Short in the Machine Learning... | Eduardo Bonet | 41m |
DevOps tooling repeatedly fails ML teams because data scientists are a different user group and ML logic is implicit in data not code; platform attempts… | ||
| Evolution and Scaling of Feature Store at Uber | Divya Nagar | 33m |
Palette serves 21,000+ features at ~25M queries/sec with 99.999% uptime and tackles scaling by automating onboarding via JSON configs, mirroring Spark and… | ||
| Using Kaskada to Do Real Time AI | Mick Semb Wever | 41m |
Kaskada is an open source Python + Rust framework (Apache Arrow for processing) that converts event timelines into features and native time travel for ML so… | ||
| A New Data Model for Time Series Data Processing | Jeff Tao | 39m |
Claims a relational variant called one table per sensor that stores time-ordered blocks, column-compresses per-device streams, and keeps tags separate so… | ||
| Building Python Binding for the Delta Lake Library in Rust | Florian Valeye | 26m |
Rust native Delta Lake library delta-rs implemented zero-copy Arrow exchange, Tokio async, and Object Store support, then published a Python extension built… | ||
| Deploying Your Own Language Models: A Use Case Demo with Open Source... | Shrey Anand, Surya Pathak | 40m |
A question-answering system for Red Hat OpenShift on AWS (ROSA) is built from public docs, indexed with Qdrant vector embeddings and served as candidates to a… | ||
| Cloud-Native MLOps Platform for Python- and ML Engineers | Joan Fontanals Martínez | 35m |
Company with ~50 people and 3+ years of history presents a cloud-native stack that moves Python ML code from local POC to Kubernetes with concrete pieces… | ||
| The Search for Transparency and Accountability in the Age of... | Ezequiel Lanza | 39m |
Argues that Responsible AI needs XAI to survive in production and shows concrete tooling and findings: uses IBM bias checks on the German credit dataset… | ||
| Expanding the Capabilities of LLMs Through Community Collaboration | Antonio Fernández, Jose García | 37m |
Community projects claim to extend LLMs beyond token limits by combining embeddings, vector stores and agents so models only receive relevant chunks; they… | ||
| Building Knowledge Graphs | Sumit Pal | 44m |
Knowledge graphs require cross‑discipline collaboration and combine technology, strategy and org change; they process 75–80% unstructured enterprise data… | ||
| Overcoming I/O Bottlenecks in LLM Training with Open-Source Distributed... | Lu Qiu, Jasmine Wang | 25m |
Alluxio implements a hierarchical distributed cache (kernel page cache → local disk pages → remote cache cluster) to cut repeated cloud reads, lower metadata… | ||
| Map of Science: A UI for Open-Source Science | Alexy Khrabrov, Tim Bonnemann | 37m |
Map of Science builds an interactive, queryable index linking people, labs, papers, datasets and OSS used in domains; a live Kumu prototype already visualizes… | ||
| nothing matches | ||