MindLedger is a decentralized academic "mining" platform. Students earn $POI tokens by solving Olympiad-level problems (JEE, INMO, IPhO). The platform uses AI-driven anti-cheat, semantic fingerprinting via pgvector, and a P2P verification network.
graph TD
User((Solver/Miner)) --> |Solve & Submit| Frontend[Next.js Application]
Frontend --> |API Requests| API[Node.js Gateway]
Frontend --> |Session Streams| WS[WebSocket Proctoring]
subgraph "Infrastructure Layer"
API --> |Context & Cache| Redis[(Redis Queue)]
API --> |Relational Data| DB[(PostgreSQL + pgvector)]
API --> |Embeddings| AI[FastAPI AI-Engine]
end
subgraph "Verification Layer"
API --> |Sandboxed Code| Docker[Docker Runner]
API --> |Solution Hash| Chain[Solidity Smart Contract]
end
Use Docker Compose to spin up the entire stack (Database, Redis, AI-Engine, Backend, and Frontend) with a single command.
Prerequisites: Docker Desktop installed and running.
-
Clone the Repository:
git clone https://github.com/anmol392/MindLedger.git cd MindLedger -
Start the Services:
docker-compose up --build -d
-
Access the Platform:
- Frontend:
http://localhost:3000 - API Server:
http://localhost:8000 - AI Engine:
http://localhost:8001
- Frontend:
Use this for granular development if you don't have Docker.
You need a running PostgreSQL (version 15+ with pgvector extension) and Redis instance.
cd backend/api_server
npm install
# Sync database schema (requires PG to be active)
npx prisma db push
# Seed initial problems
node prisma/seed.js
# Start server
node index.jscd backend/ai_engine
python -m venv venv
source venv/bin/activate # Windows: .\venv\Scripts\activate
pip install -r requirements.txt
python main.pycd frontend
npm install
npm run dev| Directory | Purpose | Tech Stack |
|---|---|---|
frontend/ |
Core user interface & dashboards | Next.js 15, Tailwind 4, Framer Motion |
backend/api_server/ |
Request gateway, proctoring & logic | Node.js, Express, Socket.io, Prisma |
backend/ai_engine/ |
Plagiarism & AI-trap detection | FastAPI, Sentence-Transformers, PyTorch |
blockchain/ |
Smart contracts & tokenomics | Solidity (ERC-20 + custom PoI logic) |
- Semantic Fingerprinting: Every solution is vectorized using
all-MiniLM-L6-v2. Any solution with >0.91 cosine similarity to existing solutions is flagged for plagiarism. - AI-Trap Injection: Problems contain subtle "hallucination traps" (physically impossible constraints) that LLMs often miss but humans catch.
- Live Proctoring: WebRTC background streams verify the presence and focus of the solver in real-time.
This project is open-source under the MIT License.
Built with precision for the global elite. © 2026 MindLedger Protocol.