ML in Production–What does “Production” even mean?
summary
Defines production as model predictions directly affecting a product or users and notes the famous 87% stat probably wrong; breaks deployment into four concrete steps — wrap trained model as a function, expose it via an API (examples: FastAPI, required /predict endpoints for some labeling tools), package it for portability (recommend Docker or equivalent and consider ONNX for portability), and host it on suitable infrastructure (cloud, on‑prem, or edge) with authentication and GPU considerations.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:54Production misunderstood
- 3:40First principles of production
- 8:17Four deployment steps
- 13:16Model deployment breakdown
- 15:47API and auth tools
- 17:53Docker containerization
- 19:18Infrastructure and GPU
- 21:03Summary and next steps
- 23:32Q&A
related talks
47mMachine Learning in Production: An Intro to MLOps.
30mThe State of Production Machine Learning in 2023
01h 05mMachine Learning Models and Productionization
21mWhy is Production ML so Hard?
44mML in Production
29mDemystifying Machine Learning in Production: Reasoning about a Large-Scale ML Platform
52mThe ML Lifecycle in Combat Conditions
25mContinuous ML Applications in Production