Machine Learning fundamentals

• Recap

  • Code fundamentals and libraries like Pandas for data manipulation

• Understanding Machine Learning

  • Machine learning fundamentals, including Supervised, Unsupervised

  • Key concepts such as features, models, training, and prediction

  • The role of machine learning in various industries and applications

• Machine Learning Algorithms

  • Machine learning algorithms: regression, classification, clustering, and neural networks

  • Mathematics and principles behind the algorithms

• Tools and Frameworks for ML

  • Machine learning tools and frameworks: TensorFlow and Scikit-learn

MLOps Project Structure and Management

• MLOps Introduction

  • The lifecycle of machine learning projects

  • The alignment of machine learning with DevOps principles

• Project Structuring

• Collaboration and Management

  • Project management tools and collaboration practices

  • Version control, issue tracking, and project documentation

• Version Control and Code Quality (GIt)

  • Git

  • Strategies, merging, conflict resolution, and pull requests

• Data Management in MLOps

  • Managing and versioning data in machine learning projects

  • The challenges related to data consistency, reproducibility, and collaboration

• Introduction to Data Version Control (DVC)

  • DVC as a solution for data versioning

  • DVC for tracking changes, sharing data, and reproducing experiments

Model Deployment and Monitoring:

• Model Deployment Strategies

  • Deployment strategies for machine learning models including Batch and Real-time

• Monitoring and Logging

  • Monitoring model performance, system metrics, and logs

  • Monitoring tools

• Understanding Batch Processing

  • Batch processing in the context of machine learning

  • Implementing Batch Jobs

  • Tools and frameworks (Apache Spark) for distributed batch processing

• Introduction to CI/CD in MLOps

  • Principles of Continuous Integration and Continuous Deployment in MLOps

  • Implementing CI/CD Pipelines

  • Testing, building, and deployment into automated workflows

  • Creatingan end to end Batch processing with CI/CD

• Real-time Processing in ML

  • Fraud detection and recommendation systems

  • The principles of real-time data streaming

  • Real-time Processing Technologies (Kafka/Flink )

• Testing in MLOps

  • Testing methodologies including unit testing, integration testing, and performance testing

  • Automated Testing Frameworks (pytest)

• Deployment and Monitoring Project

MLOps Infrastructure:

• Understanding Containers

  • Overview of containerization platforms like Docker

  • Creating and Managing Containers

  • Container Orchestration - Kubernetes

• Understanding GPU and CPU Computing

  • The role of GPUs and CPUs in machine learning computations

  • The differences, advantages, and limitations of GPU and CPU computing

  • Leveraging GPU Acceleration

  • Best practices for optimizing GPU utilization

• Introduction to Feature Stores

  • Essential in managing features for machine learning models

  • The benefits of centralizing feature management

  • Using Feature Stores  - Feast

LLM:

• LLM introduction

• LLM prompt engineering

• LLM RAG

Cloud MLOps:

• Introduction to Cloud MLOps

  • Overviw - AWS, Azure, and GCP and MLOps services

• AWS MLOps services

  • Amazon SageMaker: A fully managed service for building, training, and deploying machine learning models.

  • AWS Lambda: For serverless computing, enabling automation and integration in MLOps pipelines.

  • Amazon S3: For storing and managing data, model artifacts, and pipeline artifacts.

  • Amazon CloudWatch: For monitoring and logging of your MLOps processes and infrastructure

  • Amazon ECR (Elastic Container Registry): To store and manage Docker containers for model deployment.

• Cloud-based Machine Learning Pipelines

  • Implementing end-to-end machine learning pipelines in the cloud

  • Integratiob of various cloud services for data storage, computation, and deployment

MLOps FInal Project