A comprehensive course designed to introduce the most important areas of MLOps along with engineering best practices for building end-to-end AI solutions.
This course covers essential MLOps concepts and practices, from development to deployment. You'll learn how to build production-ready machine learning systems using modern tools and methodologies.
- MLOps introduction - dependency management, code quality, Git, FastAPI, Docker
- Databases & file formats - PostgreSQL, DuckDB, Parquet
- Data processing - Polars
- Vector databases - pgvectorscale, SQLAlchemy, Milvus
- Versioning - DVC, MLFlow
- ML testing & data-centric AI - CleanLab, Giskard, Captum, SHAP
- Model optimization for inference - PyTorch optimization, ONNX, ONNX Runtime
- Monitoring & drift detection - Evidently, NannyML
- Introduction to cloud computing - AWS services
- Infrastructure as Code (IaC) - Terraform
- Deployment & CI/CD - GitHub Actions
- ML pipelines - Apache Airflow
- LLMOps - vLLM, Model Context Protocol (MCP), guardrails
- ML system design
This course is open to everyone and is licensed under the MIT License.
This course is continuously evolving and expanding. New topics and improvements are regularly added. If you have suggestions for new topics or improvements, please feel free to open an issue.
Students and contributors who spot errors or have improvements are welcome to create pull requests or issues. Students will be rewarded with extra points for their contributions!