A production-grade AI Research Assistant built with FastAPI, Retrieval-Augmented Generation (RAG), Sentence Transformers, FAISS, BM25, Cross-Encoder Reranking, and modern AI engineering practices.
ResearchOS is an end-to-end AI-powered research assistant capable of:
- Uploading research papers (PDF)
- Extracting and processing text
- Semantic chunking
- Dense vector embeddings
- FAISS vector search
- BM25 sparse retrieval
- Hybrid Search
- Cross-Encoder reranking
- LLM-powered question answering
- Source citation support
- Multi-document retrieval
- Python 3.12
- FastAPI
- Uvicorn
- Sentence Transformers
- PyTorch
- Transformers
- FAISS
- BM25
- Cross Encoder
- Groq API
- SQLite (later milestone)
- Git
- GitHub
- VS Code
- Docker (later)
- Pytest
ResearchOS/
backend/
frontend/
docs/
notebooks/
tests/
README.md
requirements.txt
.gitignore
.env.example
- FastAPI Backend
- PDF Upload
- PDF Text Extraction
- Chunking
- Embeddings
- FAISS
- Semantic Search
- Hybrid Search
- Cross Encoder
- Chat
- SQLite
- Frontend
- Docker
- Deployment
This project is designed to learn:
- Python
- Machine Learning
- Information Retrieval
- Retrieval-Augmented Generation
- FastAPI
- Backend Engineering
- AI System Design
- Production AI Development
Every feature follows this process:
Concept
↓
Implementation
↓
Testing
↓
Documentation
↓
Git Commit
↓
GitHub Push
MIT License