Autonomous 5-stage AI research pipeline that queries PubMed's 35M+ papers, performs RAG-based literature analysis, generates novel hypotheses, designs experiments, and produces publication-style research proposals — with zero human intervention.
Topic Input
→ Stage 1: PubMed Fetch (35M+ papers, structured metadata)
→ Stage 2: RAG Indexing (FAISS + all-MiniLM-L6-v2 embeddings)
→ Stage 3: Hypothesis Gen (Groq LLaMA 3.1, top-3 semantic retrieval)
→ Stage 4: Experiment Design (dataset, methodology, metrics, timeline)
→ Stage 5: Research Report (publication-style proposal + peer critique)
ResearchMind/
├── app.py ← Gradio UI entry point
├── pipeline.py ← Core autonomous research logic
├── requirements.txt
└── README.md
git clone https://github.com/Vi-bha/ResearchMind
cd ResearchMind
pip install -r requirements.txt
export GROQ_API_KEY=your_key_here # free at console.groq.com
python app.pyOpen http://localhost:7860 in your browser.
| Component | Technology |
|---|---|
| Paper Source | PubMed API (35M+ papers) |
| Vector Store | FAISS IndexFlatL2 |
| Embeddings | SentenceTransformers all-MiniLM-L6-v2 (384-dim) |
| LLM | Groq LLaMA 3.1 8B Instant |
| UI | Gradio 4.x |
Input: large language models medical imaging
Output includes:
- 5 fetched PubMed papers (2023–2025) with metadata
- RAG-based literature summary with identified research gaps
- Novel, testable hypothesis with rationale
- Experiment design with dataset, architecture, metrics, timeline
- Peer critique with feasibility score
- Full publication-style research proposal (~1,500 words)
| Project | Description | Demo |
|---|---|---|
| PaperLens | RAG pipeline for research paper Q&A | 🤗 Live |
| MedLens | Multimodal AI for prostate MRI analysis | 🤗 Live |
Vibhavari Tummewar — M.Tech Advanced Computing, MANIT Bhopal Scopus-indexed researcher in LLM-assisted medical AI systems.
