Local-first RAG research assistant

ResearchAgent

An AI research workflow that searches arXiv, downloads academic PDFs, builds a private ChromaDB knowledge base, and answers questions with retrieval-augmented generation.

ResearchAgent local research workflow interface

AI Engineering Highlights

End-to-End RAG

Search, PDF ingestion, chunking, embeddings, vector retrieval, and grounded Gemini answers in one workflow.

Agentic Orchestration

LangGraph routes natural-language instructions to search, ingest, query, or answer tools.

Local Knowledge Base

ChromaDB, PDFs, embeddings, and fallback logs remain local by default for privacy-aware research.

Workflow

1. SearchFind papers on arXiv.
2. DownloadSave PDFs locally.
3. IngestExtract, chunk, and embed text.
4. RetrieveQuery ChromaDB for relevant passages.
5. AnswerGenerate grounded responses with Gemini.

Tech Stack

Python Generative AI LLM RAG Vector Database ChromaDB SentenceTransformers LangChain LangGraph Google Gemini Gradio PyMuPDF arXiv API Semantic Search AI Agents Observability

Product Screenshots

ResearchAgent empty agent helper state ResearchAgent agent result state