🔬 EMLOV4 dives deep into the world of MLOps, exploring advanced techniques and tools crucial for success in production environments. From Docker and PyTorch Lightning to AWS and Kubernetes, this course equips you with the knowledge and skills needed to excel in the rapidly evolving field of machine learning operations.
Note: Each of the repo may have repo secret and github workflow it was executed in seperate repos. This repo just contains all assignment solutions and they were not executed in this path or this reprository. However all the codes in this repo are executed in ajithvcoder/emlo4-session-{n}-ajithvcoder namespace where n is the session number.
Contains solutions for assignments and learning from Extensive Machine Learning Operations - Version 4.0 course of The School of AI https://theschoolof.ai/#programs
Website: [On Development]
-
Introduction to MLOps
An overview of MLOps (Machine Learning Operations), covering the best practices and tools to manage, deploy, and maintain machine learning models in production.
-
A hands-on session on creating Docker containers from scratch and an introduction to Docker, the containerization platform, and its core concepts.
Learnt about docker fundamentals and how to chose the base docker image and reduce the size as minimal as possible
-
An introduction to Docker Compose, a tool for defining and running multi-container Docker applications, with a focus on deploying machine learning applications.
Learnt about docker compose and mounting multiple volumes and handling multiple containers
-
An overview of PyTorch Lightning, a PyTorch wrapper for high-performance training and deployment of deep learning models, and a project setup session using PyTorch Lightning.
Learnt about using lighting to train, eval and infer images using a model developed.
-
Learn to build sophisticated ML projects effortlessly using PyTorch Lightning and Hydra, combining streamlined development with advanced functionality for seamless model creation and deployment.
-
Data Version Control (DVC), a tool for managing machine learning data and models, including versioning, data and model management, and collaboration features.
Medium Blogs
-
Experiment Tracking and Hyperparameter Optimization
A session covering various experiment tracking tools such as Tensorboard, MLFlow and an overview of Hyperparameter Optimization techniques using Optuna and Bayesian Optimization.
-
A session on AWS, covering EC2, S3, ECS, ECR, and Fargate, with a focus on deploying machine learning models on AWS.
-
A hands-on session on deploying machine learning models using FastAPI, a modern, fast, web framework for building APIs.
-
Gradio, an open-source platform for creating and sharing demos of machine learning models, and a session on Model Tracing.
-
Model Deployment on Serverless
An overview of Serverless deployment of machine learning models, including an introduction to AWS Lambda
-
Model Deployment w/ TorchServe
An introduction to TorchServe, a PyTorch model serving library, and a hands-on session on deploying machine learning models using TorchServe.
-
This session provides an introduction to Kubernetes, a popular container orchestration platform, and its key concepts and components.
-
In this session, participants will learn how to monitor and configure Kubernetes clusters for machine learning workloads with knowledge on Helm.
-
Kubernetes-III ALB - AutoScaling
This session will cover introduction to EKS, Kubernetes Service on AWS, Deploying a FastAPI - PyTorch Kuberentes Service on EKS, ALB, Cluster auto scaling, HPA
-
This session covers EBS Volumes, ISTIO and KServe, learning to deploy pytorch models on KServe
-
Canary Deployment & Monitoring
This session covers how to deploy models with Canary Rollout Strategy while monitoring it on Prometheus and Grafana
-
This session is a final project where participants will apply the knowledge gained throughout the course to develop and deploy an end-to-end MLOps pipeline.
Important tools, method, configs, links
- Add in bio website with all repo links
-
Every month end during course development
-
There is a resource file in which i maintain that has concepts and tools which i learnt newly in EMLO-4.0 course.
Thanks to Satyajit Ghana for this course.
I really worked very hard to complete each of the assignments.
Thus the course was completed
