This project was made as part of the GydeXP Hackathon (in collaboration with the AI/ML Society of IIT Delhi), where our team was selected as Finalists among the top 14 teams at the National Level (India).
Team Members:
- Urvi Agrawal
- Rutuja Nimje
- Simran Hedaoo
Welcome to the Digital Welfare Companion 👋
This is an AI-powered agentic system built with CrewAI, Streamlit, and Gemini API to make government schemes accessible in local languages.
git clone https://github.com/yourusername/Resume-Screening-Crew.git
cd resume_checkerpip install -r requirements.txtCreate a .env file and add your API credentials:
GEMINI_API_KEY="your_api_key"
streamlit run streamlit_app.py- Home Page – User lands on a simple, multilingual interface with voice and text query options.
- Voice/Text Query Processing – Input is sent to Google Gemini API for scheme discovery and interpretation.
- Agentic Scheme Matching – CrewAI agents identify relevant government schemes, extract eligibility, documents, and application steps.
- Document Upload & OCR – Users can upload Aadhaar, income certificate, etc., which are parsed via Pytesseract OCR.
- AI-based Document Verification – Agents check compliance of uploaded documents with scheme requirements.
- Structured Response Generation – A personalized, step-by-step report is generated with eligible schemes, required documents, and next actions.
- Frontend/UI – Streamlit with Custom CSS for an accessible and simple user experience
- Voice Input & Query Handling – Streamlit + Google Gemini API for natural language understanding
- Multi-Agent System – CrewAI for distributed reasoning and task delegation
- OCR & Document Processing – Pytesseract (Tesseract OCR) for extracting text from uploaded documents
- APIs/Integrations – Google Gemini API for scheme discovery & language support
- Deployment Ready – Streamlit app structure for easy hosting and demoing