TRACK-1_MAVERICKS is a TypeScript-based project built with Vite, Supabase, and modern tooling for rapid development.
This repository is part of the VectorFlow-vvce initiative and is designed to provide a scalable foundation for building web applications.
Eliminating the information gap between Indian farmers and ₹1.5 Lakh Crore in unclaimed welfare subsidies.
Indian farmers lose thousands of crores annually to information asymmetry and middlemen exploitation. Kisan Saathi is a personalized "Welfare Intelligence Engine" that uses local AI and voice-recognition to match farmers directly to their eligible government subsidies—removing corruption and friction from the application process.
- 🗣️ Voice-First Vernacular Interface: Farmers can describe their situation in Kannada, Hindi, or Tamil. Powered by OpenAI Whisper, ensuring accessibility for users with limited digital literacy.
- 🛠️ Kiro-Driven Spec Architecture: Designed using Spec-Driven Development in Kiro to ensure mathematically accurate matching logic for complex eligibility criteria.
- 🧠 Edge-AI Intelligence: Runs a local Ollama + Mistral 7B model. This provides expert financial advice completely offline, ensuring privacy and reliability in rural low-connectivity zones.
- 🛡️ Anti-Bribe Roadmap: Generates a personalized "Document Checklist" and "Transparency Pipeline" so farmers know exactly what they need and where their application stands.
The system utilizes a RAG (Retrieval-Augmented Generation) pipeline to ensure all advice is grounded in current government law.
- Ingestion: User inputs land size, crop type, and location via voice or text.
- Kiro Engine: Processes input into vector embeddings for precise matching.
- Knowledge Retrieval: Queries a verified database of 1500+ government scheme JSON files.
- Local LLM Analysis: Mistral 7B predicts eligibility and calculates the "Financial Opportunity" for the user.
- Multilingual Output: Results translated and displayed on a clean, high-fidelity Streamlit dashboard.
- src/ → Main application source code
- supabase/ → Supabase configuration and migrations
- .env → Environment variables (not committed)
- package.json → Dependencies and scripts
- vite.config.ts → Vite configuration
- tsconfig.json → TypeScript configuration
- eslint.config.js → Linting rules
- wrangler.jsonc → Cloudflare Workers configuration
- bun.lockb / bunfig.toml → Bun runtime configuration
- TypeScript (98%)
- CSS (1.7%)
- JavaScript (0.3%)
- Supabase for backend services
- Vite for fast bundling
- Bun as runtime and package manager
- Cloudflare Workers for deployment
- Clone the repository
git clone https://github.com/VectorFlow-vvce/TRACK-1_MAVERICKS.git cd TRACK-1_MAVERICKS
pip install -r requirements.txt
streamlit run app.py