This repository provides a free, 10-week online course focused on building AI systems for knowledge base question answering using Large Language Models (LLMs). It targets data engineers, ML engineers, and researchers seeking practical, hands-on experience with LLMs, Retrieval-Augmented Generation (RAG), vector databases, and system evaluation. The course offers a structured curriculum with video lectures, homework, and a capstone project to build real-world AI applications.

How It Works

The course follows a modular approach, starting with LLM fundamentals and RAG, then progressing to open-source LLMs (Ollama, GPU deployment), vector databases (embeddings, indexing), evaluation, monitoring, and orchestration (Mage). It emphasizes practical implementation, covering topics like hybrid search and document reranking, culminating in an end-to-end project. The curriculum is designed to equip learners with the skills to build and deploy functional AI-powered Q&A systems.

Quick Start & Requirements

• Self-Paced Learning: Watch videos, complete homework, and work on a project.
• Prerequisites: Basic understanding of data science and Python. Specific module requirements (e.g., GPU for certain deployments) may apply.
• Resources: Links to course videos, homework, and a community Slack channel are provided.

Highlighted Details

• Comprehensive 10-week curriculum covering LLMs, RAG, vector search, evaluation, and monitoring.
• Hands-on experience with tools like OpenAI API, Elasticsearch, Ollama, and vector databases.
• Focus on building end-to-end AI systems for knowledge base question answering.
• Includes bonus modules on advanced techniques like hybrid search and document reranking.

Maintenance & Community

• Active community support via a dedicated Slack channel (#course-llm-zoomcamp).
• Organized by DataTalks.Club, a global online community for data enthusiasts.
• Links to Telegram announcements, course playlist, and community guidelines are available.

Licensing & Compatibility

• The course content and associated code are generally provided for educational purposes. Specific licensing for code snippets or datasets should be verified within the repository.

Limitations & Caveats

• The course is structured for specific cohorts (e.g., June 2, 2025), but self-paced learning is supported. Some advanced topics might require specific hardware (e.g., GPUs) for optimal local execution.