A Deep Dive into Machine Learning Pipeline with ML Ops architecture, Model registry, and Feature Store

Following are key functionalities into a modern, scalable and reliable machine learning pipeline along with model deployment, prediction pipeline, and various other functionalities:

1. ML Training platform with AutoML, stacking, blending
2. ML Observability and monitoring
3. Model Registry
4. Model Transformation
5. Feature Store

For any enterprise, a reliable and predictable ML pipeline is the most important aspect of their MLOps and DevOps teams. In this video we are going to go deep into the machine learning pipeline and understand some of the key concepts:

• (00:00) Video starts
• (00:08) Content intro
• (03:26) Deep Dive Starts
• (04:28) Data Preparation and Pre-processing
• (08:55) Machine Learning Frameworks
• (13:00) ML Training Frameworks
• (15:56) Statistical ML and Deep Learning
• (20:00) AutoML & Blending/Stacking
• (27:56) ML Ops Platforms
• (31:15) MLOps Architecture
• (36:00) Model Registry
• (40:07) Model Transformation (ONNX)
• (43:12) Data, Code, and Model Artifacts
• (47:42) Feature Store
• (55:05) Recap

Thanks for your time, Avkash Chauhan