Applied AI · retrieval
Resume RAG.
A resume question-answering application using vector retrieval and context-constrained responses.
in context.Original architecture illustration · source-based
01 / The context
A problem worth solving.
Resume questions should be answered using the document's content, with a clear response when the context does not contain the requested information.
02 / My contribution
The engineering underneath.
Connected a TypeScript React client to a Python FastAPI service. The LangChain pipeline retrieves the top three similar resume chunks from ChromaDB and supplies them as answer context.
Implemented features
- React and TypeScript client
- FastAPI service and LangChain pipeline
- Similarity retrieval over resume chunks
- Explicit unavailable-information response
Inside the system
Follow the thinking.
Explore a source-based interaction study. No live data, AI calls, or transactions.
Ask a question
The React client sends a resume question to a FastAPI service. This is a source-based explanation, not a running AI endpoint.
Illustration of documented behavior
Read the complete flow as text
- Ask a question · QUESTION
The React client sends a resume question to a FastAPI service. This is a source-based explanation, not a running AI endpoint.
- Retrieve context · RETRIEVAL
LangChain retrieves the top three similar resume chunks from ChromaDB for the response context.
- Answer within scope · RESPONSE
The prompt constrains answers to retrieved context and includes an unavailable-information response. This does not guarantee hallucination prevention.
04 / Scope & boundaries
An honest account of the scope.
Source-verified application. No live deployment or measured accuracy is claimed; context constraints do not guarantee hallucination prevention.
Read the implementation.
The repository contains the source and supporting context for this project.
View on GitHub (opens in new tab)