Building a secure MLOps pipeline, featuring PathAI
summary
Secure MLOps pipeline on AWS uses SageMaker plus KMS, S3 Object Lock, Macie, GuardDuty, Inspector and Security Lake to protect data and detect threats, with CI/CD promoting models across research / staging / production accounts and Amazon SageMaker Model Registry and Pipelines for versioning. PathAI packages models into containers, ships them to the AISight platform, and monitors with SageMaker Model Monitor; outcome: 1.5M+ WSIs processed, 24M annotations, 250k training experiments, 200+ models in production, faster prototypes and zero critical high‑risk exploits reported.
AI-written summary · about
chapters (11)
- 0:00Introduction to MLOps
- 2:25ML SecOps Framework
- 7:52AWS MLOps Security Tools
- 12:08Data & Development Security
- 21:11Model Registry & Deployment
- 24:28Access Control Strategies
- 25:06Model Monitoring & Pipelines
- 30:22Generative AI MLOps
- 31:48Case Study: Path AI
- 43:20Path AI Architecture & Results
- 47:56Key Takeaways & Resources
related talks
26mRunning Secure AI/ML Experiment in a Unified MLOPS-DevSecOps Pipeline
34mImplement MLOps practices with Amazon SageMaker
37mStreamlining Machine Learning Operations
43mMachine Learning Ops: DevOps for Data Science
29mImplementing MLOps practices with Amazon SageMaker
34mImplement MLOps practices with Amazon SageMaker
33mBuilding & Managing MLOps Pipelines in the Cloud
36mImplementing MLOps practices on AWS