Advanced · 4–5 weeks
RAG-Powered Internal Knowledge Base
Ingest company docs (Markdown/PDF), chunk and embed into vector DB, expose chat API that retrieves relevant context and answers with citations. Admin UI to re-index documents.
Why this project
RAG is the most common production AI pattern. Full-stack Java implementation with Spring AI is highly relevant in 2025–2026.
Tech stack
Java 17Spring Boot 3Spring AIOpenAI/Anthropic APIpgvector or PineconePostgreSQL
Skills demonstrated
- RAG pipeline
- Chunking strategies
- Vector search
- Citation grounding
Interview talking points
- Chunk size trade-offs
- Handling stale documents
- Eval metrics you used
- When RAG fails and fallback
Problem statement
Ingest company docs (Markdown/PDF), chunk and embed into vector DB, expose chat API that retrieves relevant context and answers with citations. Admin UI to re-index documents.