Backend scaffold for a multi-stage scientific reasoning platform.

• Python
• FastAPI
• PyTorch
• Transformers
• Neo4j (graph storage)

• app/main.py: FastAPI application entrypoint
• app/api.py: HTTP routes and endpoints
• app/schemas.py: Pydantic request/response models
• app/services/: Domain logic for document parsing, claim extraction, multi-agent review, and graph building

1. Create a Python environment
2. Install dependencies from requirements.txt
3. Run the app:

uvicorn app.main:app --reload --port 8000

• Implement parser integration with GROBID / Science Parse
• Add SciBERT claim extraction and LLM verification
• Build the multi-agent review orchestration layer
• Connect the knowledge graph builder to Neo4j

Run Instructions

Prereqs: Ensure Node+npm and Python 3+ are installed. Optional: Neo4j if you want graph features.

Backend (Windows PowerShell):

(If needed) create venv and install deps:

python -m venv .venv ..venv\Scripts\python.exe -m pip install -r requirements.txt

Start the API: .venv\Scripts\python.exe -m uvicorn app.main:app --host 127.0.0.1 --port 8000

Frontend:

cd frontend npm install # only if dependencies not installed npm run dev

Dev server URL: http://localhost:4173/

Alternative (Docker Compose): docker-compose up --build