Open AI + Data Forum @ Open Source Summit North America 2023
2023-05-10..2023-05-12
-- online --
[Open AI + Data Forum @ Open Source Summit North America 2022, Open AI + Data Forum @ Open Source Summit North America 2023, Open AI + Data Forum @ Open Source Summit North America 2024]
who it's for
data engineers, machine learning engineers, mlops engineers, data scientists, ai researchers, platform engineers, security engineers
topics
Delta Lake time travel; LakeFS git-like data versioning; torch.compile in PyTorch 2.0; ModelMesh multi-model serving; federated learning with OpenFL and FATE; homomorphic encryption implementations; vector databases (Milvus); Presto SQL engine for lakehouses; CUTLASS CUDA template library; xBOM eXtensible bill of materials; feature ablation for explainability; streaming data applications
| title | author | length |
|---|---|---|
| Socialising the Elephant with the Rest of the Animals - Integrating... | David Radley, Jürgen Hemelt | 50m |
Egeria provides an open, peer-to-peer metadata layer that maps vendor schemas to a common type system and runs in production at IBM/Atruvia to give a single… | ||
| Getting ML Right in a Complex Data World | Vinodhini Duraismy | 31m |
A data versioning engine called LakeFS provides git-like data versioning on top of S3-backed object stores (also works with MinIO), letting teams branch, tag… | ||
| Accelerate Model Training with an Easy to Use High-Performance... | Michael Clifford, Erik Erlandson | 45m |
A reproducible stack is demonstrated that lets data scientists declare resources in Python and push distributed jobs without DevOps: CodeFlare SDK + operator… | ||
| Getting Started with Delta Lake | Guenia Izquierdo Delgado, Sajith Appukuttan | 38m |
Delta Lake adds an ACID layer over cloud object storage, storing a Delta log of JSON/parquet commits so tables support time travel and consistent concurrent… | ||
| Fostering Trust via Explainable ML Inferences | Dalmo Cirne | 41m |
Claims a practical, quantifiable game‑theory framework that turns explainable inferences into measurable trust: use feature ablation (remove fields to see… | ||
| Contributing AI-Generated Content to OSS Projects - Navigating Legal, Copyright, &... | Joanna Lee | 38m |
AI output itself is not copyrightable but human edits are, and AI can still infringe third‑party rights when its training data is reproduced; pending… | ||
| Fairness in Machine Learning: From Theory to Practice | Alex Karsten | 42m |
Talk argues that perfect fairness is impossible but actionable steps reduce harm: measure subgroup performance using sensitivity and specificity, pick… | ||
| What's New, ModelMesh? Model Serving at Scale | Rafael Vasquez | 36m |
KServe’s ModelMesh serves as a Kubernetes multi-model backend that hosts 60,000 models managed while keeping only about 5,000 in memory, using a distributed… | ||
| Weaving the [X]BOM Fabric | Erica Dretzka | 44m |
Proposes an eXtensible Bill of Materials pattern that digitizes hardware, software, data and people into interoperable BOMs and pilots xBOM Fabric prototypes… | ||
| PyTorch 2.0: Unlocking the Power of Deep Learning with the Torch Compile API | Christian Keller | 15m |
PyTorch 2.0 introduces torch.compile as a one-line API that combines Dynamo, AOTAutograd and torch.inductor to capture graphs, lower ~2–3k ops to ~200–300… | ||
| Federated Learning: Unlocking the Values of Discrete Data for Enterprise | Henry Zhang, Layne Peng | 42m |
Overview of production federated learning that cites two open source stacks—FATE and OpenFL—and tooling to run them across clouds including CubeFate and… | ||
| Understanding the Legal Context of AI | Van Lindberg | 47m |
Recent lawsuits name GitHub Copilot, Stable Diffusion, DALL‑E, Midjourney and the legal analysis ties directly to how models are built: models are matrices of… | ||
| OpenFL: A Federated Learning Project to Power (and Secure) Your Projects | Ezequiel Lanza | 32m |
OpenFL is a Python 3 framework for federated learning that actually ships: a pip package, containers, workspace/plan files and an integrated certificate… | ||
| How to Eliminate the I/O Bottleneck and Continuously Feed the GPU While Training in the... | Lu Qiu | 27m |
Describes Alluxio’s new Dora decentralized repo sitting between compute and cloud storage, built as worker + server-registry + client with no single point of… | ||
| Homomorphic Encryption in the Open Source World | Aakanksha Duggal | 33m |
A Red Hat research effort compares open libraries and builds proofs of concept: Microsoft SEAL is the C++ leader, older Paillier-style code is ~1200× slower… | ||
| The Evolution of the Open Data Ecosystem | Ranadip Chatterjee | 46m |
Tracing Hadoop from MapReduce (2004) to Hadoop clusters (2006–07), the claim is that open source solved vendor lock-in but created a spaghetti of components… | ||
| 5 Steps to Deploy Cloud Native Sustainable Foundation AI Models | Chen Wang, Huamin Chen | 48m |
Shows how to containerize foundation models and run Bloom (516M) and Stable Diffusion in Docker on Kubernetes, instrumented with NVIDIA DCGM exporter +… | ||
| A MLOps Platform for All Data Personas | Karan Honavar | 52m |
IKEA builds a custom MLOps platform that standardizes seven data personas, moves from MVP to MSP toward a target of 200–300 models, and runs mixed cloud plus… | ||
| Object Storage Driven Machine Language Workloads | Daniel Valdivia | 24m |
Argues that designing pipelines around object storage removes painful data transfers and enables scalable, auditable ML workflows; demonstrates deploying MinIO… | ||
| Transformes for Time Series: Is the New State of the Art (SOA) Approaching? | Ezequiel Lanza | 54m |
Claims that transformers beat LSTMs on long-horizon forecasting for time series get tested with open source Informer and SpaceTimeFormer code, using… | ||
| Relational Observability for Cloud-Native Security and Data Science | Frederico Araujo, Teryl Taylor | 49m |
SysFlow captures eBPF system calls, transforms them via libsysflow and a sysflow-collector into an object-relational format, and the stream processor builds… | ||
| Presto Open Source SQL Engine for Massive Scale Data Lakehouses | James Cho | 51m |
Presto is a disaggregated SQL engine that queries data in place (object storage, Hive metastore, RDBMS) with a coordinator + worker model that scales to… | ||
| AI/ML Data Pipeline Processing with Go Microservices Based... | Neethu Elizabeth Simon, Samantha Coyle | 43m |
Containerized Go microservices orchestrate automated image transfer, preprocessing and inferencing using EdgeX Foundry with Redis for job/task state, Project… | ||
| Introducing Prestissimo, Presto's Next-Gen Query Execution Engine | Aditi Pandit | 35m |
Prestissimo is a drop-in C++ worker for Presto that replaces the Java worker and runs plans over the Velox vectorized runtime. Benchmarks on Parquet data in S3… | ||
| Building Modern Data Streaming Apps with Open Source | Timothy J Spann | 47m |
Shows a working pipeline that ingests REST APIs, transit feeds and ADS‑B aircraft data with Apache NiFi, publishes to Kafka or Apache Pulsar, runs continuous… | ||
| CUTLASS: A CUDA C++ Template Library for Accelerating Deep Learning... | Aniket Shivam, Vijay Thakkar | 35m |
CUTLASS is an open source header-only CUDA C++ template library that implements high-performance GEMM and convolution kernels since 2017, claims parity with… | ||
| Xtreme1, the Next GEN Platform for Multisensory Training Data | Lin Du | 15m |
Xtreme1 is an open-source multimodal training-data platform (GitHub) hosted in the Linux Foundation AI & Data landscape that claims a 72% efficiency boost for… | ||
| Using Apache OpenNLP with OpenSearch K-NN Vector Search | Jeff Zemerick | 20m |
Shows a working pipeline that exports a pre-trained sentence-transformers model to ONNX export, calls it from Java via ONNX Runtime inside Apache OpenNLP to… | ||
| The Rise of Vector Databases - Lessons from the Milvus Community | Frank Liu, Charles Xie | 26m |
Explains why Milvus is a purpose-built vector database (open source, ~20,000 stars on GitHub) and when to use it versus raw libraries like FAISS, showing that… | ||
| MindSpore's Foundation Model Platform Promotes the Ecological Development of Foundat... | Yuanke Zhong | 24m |
MindSpore is an open source AI framework that natively supports training huge foundation models and now runs a public platform that provides a model library… | ||
| nothing matches | ||