Back to Projects
AI
YDKJS AI Tutor
Full-stack RAG-powered reading assistant for the 'You Don't Know JS' book series. Features semantic search and contextual Q&A using local LLMs.
About This Project
This project represents my exploration into practical AI engineering, combining my love for JavaScript education with modern AI capabilities.
The YDKJS AI Tutor is designed to help developers truly understand the nuances of JavaScript by providing contextual explanations and answering questions based on Kyle Simpson's acclaimed book series.
The system uses Retrieval-Augmented Generation (RAG) to ground responses in the actual book content, ensuring accurate and relevant answers rather than generic AI responses.
Key Features
- Semantic search across all YDKJS books
- Contextual Q&A with source citations
- Local LLM support via Ollama
- Vector embeddings with ChromaDB
- Dockerized for easy deployment
Technical Challenges
- Optimizing chunk sizes for accurate retrieval
- Balancing response quality with latency
- Handling context window limitations
Tech Stack
Next.jsFastAPIPythonOllamaChromaDBDocker
Impact
Personal learning tool with RAG architecture