DataEngBytes 2023 - SYD-T2-03 - Akanksha Malik
summary
Practical checklist for turning prototypes into production: define team roles (data engineer, data scientist, ML engineer), choose Dev/Test/Prod or limited-access multi‑environment, and treat regions and regulatory constraints as first‑class requirements. Implement four automated pipelines — data pipeline, environment pipeline, model training pipeline, and deployment pipeline — with clear triggers (ingest hourly, retrain weekly/monthly), rollback to last registered model, and continuous monitoring for data drift, model validation, naming, docs, cost and compute emissions. Use cloud-native tooling (Azure examples) and automated rollbacks to avoid downtime.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 1:05MLOps considerations matter
- 2:30Model to production journey
- 4:57MLOps principles
- 7:23MLOps process and roles
- 9:07Team structure and setup
- 11:36Environment strategies
- 14:46Regions and governance
- 17:12Automation pipelines
- 20:59Rollbacks and monitoring
- 21:41MLOps best practices
- 23:48Cloud environments
- 24:30Recap and start MLOps
related talks
28mDataEngBytes 2023 - PER-T1-05 - Ellie Gholami
27mDataEngBytes 2023 - PER-T1-08 - Sergio Zenatti Filho & Elizabeth Antoinne
39mDataEngBytes 2023 - BRI-T1-02 - Mikiko Bazeley
23mDataEngBytes 2023 - MEL-T3-01 - Rajath Vanikul
39mDoing Data: The Critical Process of Data Preparation (...)
31mEngineers guide for shepherding models in to production
17mProductionizing Data Science Workloads
41mData engineering without walls