Interview
summary
Describes replacing ad hoc dataset blobs with a table abstraction inside a multitenant data lake so ML pipelines can get consistent, point‑in‑time features; implementation uses a transaction‑log‑first backing to enable time travel and reproducible training, plus a lightweight API for tenant isolation. Reports that this change cuts debugging and data drift errors that previously broke models, explains the migration mistakes encountered, and shows the new flow running in production.
AI-written summary · about
chapters (4)
- 0:00Intro
- 0:32What is your talk about
- 1:35What is your role at Salesforce
- 2:49What are you seeing in the data space
related talks
54mThe Future of Data Science/ML at Scale: A Look at MLflow, Delta Lake + Lakehouse
34mData Lakehouse: Who, Why and How
01h 03mReinvente su estrategia de datos con Lagos de Datos Transaccionales
45mThe Data Lakehouse: A tech talk for everyone who loves data
51mSolving the challenges of adding transactional support to Data Lakes
31mData Warehouses Meet Data Lakes
34mTales of Data Architecture Evolution
41mModern Data Architectures – What They Are and How to Implement?