MLOps

Practical in depth hands-on MLOPs utilising best available tools [known in 2022], practice and strategy.

01 Intro to MLOps, Initial Setup, and Docker

02 Hydra, Project Templates

03 DVC - Data Version Control and Experiment Tracking

Reference implemetation in repo https://github.com/aiplaybookin/lightning-hydra-template

04 Deployemnt for Demos

Gradio App (or Streamlit)
APIs (testing using hoppscotch)
Torch Script vrs Trace

05 AWS - Training & Deployments

EC2
S3
ECS
ECR
Spot Instances, EKS, Lambda, Kinesis, Firehose, Sagemaker

06 Distributed Training and Case Study

07 Model Explainability

08 Model Serving with Torch Serve

09 Deployment on Accelerators (AWS Inf) and Serverless Inference

10 Deployment on Edge Devices (Jetson nano)

11 Model and Data Drift

12 AWS Serverless Best Practices

13 Kubeflow, Sagemaker Pipelines and Kafka

14 CI/CD Pipeline with AWS/Jenkins