Course Wiki

LLM Zoomcamp

DataTalks.Club's free ten-week course on building production LLM applications: agentic RAG, vector search, evaluation, monitoring, hybrid search, and reranking.

LLM Zoomcamp is DataTalks.Club’s free ten-week course on building practical, production-ready LLM applications. It covers Retrieval-Augmented Generation, vector search, embeddings, AI agents, function calling, evaluation, monitoring, hybrid search, and reranking, and ends with a capstone: a complete RAG application the learner designs, builds, evaluates, and ships. No ML background, GPU, or expensive setup is required — confident Python, command-line comfort, basic Docker familiarity, and a few dollars of API credits cover it.

All materials are open source in the course repository, with videos on YouTube and a course FAQ.

Curriculum

All modules, each with its lesson-level notes:

• Module 1: Agentic RAG
• Module 2: Vector Search
• Module 3: Orchestration
• Module 4: Evaluation
• Module 5: Monitoring
• Module 6: Best Practices

• Module 7: End-to-End Project

• Module 1: Agentic RAG — build a RAG pipeline with keyword search, add embeddings and vector search, then make it agentic with function calling (RAG, agentic RAG).
• Module 2: Vector Search — semantic search with embeddings across minsearch, sqlitesearch, and PGVector.
• Module 3: AI Orchestration — LLM workflows as Kestra orchestrations (workflow orchestration).
• Workshop: Data Ingestion — dlt pipelines for ingesting and analyzing LLM traces, with filesystem and REST API sources, DuckDB, and marimo dashboards.
• Module 4: Evaluation — measuring retrieval and answer quality, offline and online (LLM evaluation).
• Module 5: Monitoring — user feedback, system health, and live dashboards (LLM monitoring).
• Module 6: Best Practices — LangChain, hybrid search combining vector and keyword retrieval, and reranking for precision.
• Module 7: End-to-End Project Example — a complete fitness assistant built with LLMs.
• Capstone — a searchable knowledge base, retrieval pipeline, evaluation process, user-facing interface, and monitoring loop, all learner-owned.

Looking for a specific concept? The Concept Glossary links every concept this course teaches to the exact modules and lessons that cover it.

Who it is for

The course targets software engineers adding LLM and search capabilities to real products, data engineers wiring retrieval pipelines into production systems, and ML practitioners who want a structured way to evaluate and monitor LLM-based applications. AI Dev Tools Zoomcamp is the complement for the software-development workflow around such projects rather than the application itself.

Projects and certificate

The capstone requirements mirror what production RAG work actually needs: ingest and store a chosen dataset, implement the full retrieve-assemble-call flow, measure retrieval and answer quality with search metrics or LLM-as-a-judge, expose a UI or API, and track queries and feedback over time. Past cohort projects include fitness and nutrition assistants, study companions, medical FAQ assistants, codebase Q&A bots, and news summarization tools. Certificates require the capstone plus peer reviews of three peers’ projects during a live cohort.