Registry Synced

BSDA5014 - Machine Learning Operations (MLOps)

657 words
3 min read
FieldValue
Course CodeBSDA5014
LevelDegree Level Course
Credits4
TypePre-requisites:
Pre-requisitesBSCS1002 -  Programming in Python

📖 Description

This course aims to give students a comprehensive understanding of Machine Learning Operations (MLOps). MLOps is a paradigm to deploy and maintain machine learning models in production environments reliably and efficiently. The course will cover various aspects of MLOps, including model development, model deployment, monitoring, and optimization. Students will gain hands-on experience with popular MLOps tools and frameworks, enabling them to manage machine learning workflows effectively to deliver robust and scalable ML solutions.

🗓️ Weekly Syllabus

WeekTopic
Week 1Introduction to MLOps: Overview of MLOps and its significance: Key challenges in deploying and managing ML models in production, Comparison of traditi
Week 2ML Pipelines & Data Management: Overview of data engineering tools and practices, Data management for ML models, ML pipeline automation. DVC overview
Week 3Data Management - Part 2 : Feature Stores. Motivation, role in ML and Generative AI applications, benefits for MLOps. Feast overview.
Week 4CI/CD for ML Models: Use of version control systems like Git for model development, automated testing & validation, model delivery strategies
Week 5Machine Learning Model Development: Comparison of development of small models vs large models including LLMs, tracking model training & experimentatio
Week 6Model Deployment and Serving: Overview of containerization and orchestration technologies and various other cloud-based deployment options, Deployment
Week 7Monitoring and Performance Optimization: Techniques for monitoring model performance in production, Logging and error tracking for ML systems, Perform
Week 8ML Security: Overview of Security considerations for ML, field of MLSecOps, tooling options.
Week 9ML Governance: Overview of model explainability and ethical considerations in ML deployments, tracking bias
Week 10MLOps for LLMs - Part 1 : Model Versioning for base models and fine-tuned variants, CI/CD specifics
Week 11MLOps for LLMs - Part 2 : Accuracy, Performance vs Cost tradeoffs, Observability, Security & Governance (Bias, Toxicity, Explainability)
Week 12Closing topics: Advanced topics of Federated learning, edge inferencing; Putting it all together

📚 Books & Resources

Prescribed Books The following are the suggested books for the course:
        Building Machine Learning Powered Applications: Going from Idea to Product - Emmanuel Ameisen - O’Reilly publication
        
        Reliable Machine Learning: Applying SRE Principles to ML in Production - Chen, Murphy, Parisa - O’Reilly publication
        
        Machine Learning Engineering in Action - Ben Wilson - O’Reilly publication
        
        Designing Data-Intensive Applications: The Big Ideas Behind Reliable, Scalable, and Maintainable Systems - Martin Kleppmann - O’Reilly publication

📝 About the Instructors

Rangarajan Vasudevan
Co-Founder & Chief Data Officer
,
Lentra.ai
Rangarajan Vasudevan is the Co-Founder & CDO of Lentra.ai, India’s fastest growing lending cloud. He did “big data” & “data science” before it was fashionable, building data-native applications across industries and geographies over 15+ years.
...
more
Ranga joined Lentra by way of an acquisition in June 2022 of his company TheDataTeam, creators of Cadenz.ai customer intelligence platform. Prior to founding TheDataTeam, Ranga served as Director, Big Data with Teradata Corporation’s international business unit. Ranga joined Teradata via the acquisition of Aster Data Systems, where he was a founding engineer and co-invented a company-defining, patented, pattern recognition algorithm. He is a recipient of both the Distinguished Engineer (R&D) and Consulting Excellence awards while at Teradata.
Ranga has degrees in Computer Science from the University of Michigan and IIT Madras.
Faculty mail : rangarajan[at]study[dot]iitm[ac]in
less
Other courses by the same instructor:
BSDA5001 -
Introduction to Big Data

Document Outline
Table of Contents
System Normal // Awaiting Context

Intelligence Hub

Navigate the knowledge graph to generate context. The Hub adapts dynamically to surface backlinks, related notes, and metadata insights.