Open Source Summit Latin America 2022: Open AI + Data Forum
2022-08-23..2022-08-24
-- online --
[Open Source Summit Latin America 2022, Open Source Summit Latin America 2022: Open AI + Data Forum]
[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 Europe 2023, Open Source Summit Europe 2022: Open AI + Data Forum, Open AI + Data Forum @ Open Source Summit North America 2024, Open Source Leadership Summit 2018]
who it's for
machine learning engineers, MLOps engineers, data engineers, platform engineers, kubernetes operators, security engineers, analytics engineers
topics
fake news NLP classification; TensorFlow Serving on Kubernetes; TFX and Kubeflow pipelines; federated learning with TensorFlow Federated; model monitoring without labels (NannyML); adversarial robustness and fairness tooling; SBOMs and supply chain security; Apache Pinot real-time analytics; Apache Druid high QPS analytics; object storage backed ML pipelines
| title | author | length |
|---|---|---|
| Applying Machine Learning Techniques to Detect and Understand Fake News | João Lucas Vale Assis | 34m |
A machine learning pipeline scrapes Brazilian news (G1) and fact checks (watos.org), uses Python, Jupyter, pandas, BeautifulSoup, NLTK, TF‑IDF and Word2Vec… | ||
| Deploying ML at Scale with Kubernetes and TFX | Rishit Dagli | 42m |
Deploys a SavedModel (ResNet ~110 MB stored on GCS) with TensorFlow Serving inside Kubernetes, exposing REST on port 8501 and gRPC endpoints, serving a sample… | ||
| Federated Machine Learning with Kubernetes | Rishit Dagli, Shivay Lamba | 31m |
A concrete demo shows federated training of a Keras MNIST-style model wrapped with TensorFlow Federated where 10 simulated clients each train locally and send… | ||
| How to Detect Silent Failures in ML Models | Wojtek Kuberski | 22m |
Open-source Python library NannyML implements a concrete method: it calibrates model probabilities, applies a decision threshold, builds per-row partial… | ||
| Keeping Networks Secure: The Open Way | Ranny Haiby | 28m |
Argues that open source accelerates 5G standards by creating a feedback loop between implementations and specs and shows concrete security practices: industry… | ||
| MLSecOps with Automated Online and Offline ML Model Evaluations on Kubernetes | Tommy Li, Animesh Singh | 40m |
Open source tools run automated adversarial and fairness checks during training and at runtime using Adversarial Robustness Toolbox and AI Fairness 360, with… | ||
| Real-Time Analytics: Going Beyond Stream Processing with Apache Pinot | Karin Wolok, Mark Needham | 34m |
A working pipeline ingests the Wikimedia recent-changes stream into Python, writes to Kafka, and consumes into Apache Pinot with a Streamlit dashboard that… | ||
| Tales at Scale: Analytics at 1000 QPS and Beyond | Gian Merlino | 27m |
Explains how Apache Druid achieves thousands QPS with concrete techniques: columnar segmented storage with dictionaries and indexes that avoid reading unused… | ||
| Training AI To Code Using The Largest Code Dataset (Project CodeNet) | Christian Kadner, Tommy Li | 39m |
Project CodeNet provides 14 million submissions across 4,000 problems and 55 languages (over 500 million lines of code) with per-submission CPU/memory and… | ||
| Writing Machine Learning Pipelines Against Object Storage | Daniel Valdivia | 35m |
Argues for building ML pipelines that read and write directly to object storage (MinIO) so compute only pulls batches instead of copying entire datasets, using… | ||
| nothing matches | ||