title	TalentScout AI
emoji	🤖
colorFrom	blue
colorTo	green
sdk	streamlit
sdk_version	1.27.0
app_file	frontend/app.py
pinned	false

TalentScout AI Hiring Assistant

An intelligent chatbot for technical candidate screening — built with Python, Streamlit, FastAPI, Groq/Llama, and Qdrant.

🚀 Features

• Friendly Chat UI: Modern, multi-step, streaming chat interface (Streamlit)
• Resume PDF Parsing: Extracts name, contact info, tech stack, and experience
• LLM-Driven Interview: Interactive, context-aware interview using Groq (Llama3-70B)
• Auto-Generated Tech Questions: Dynamic based on declared tech stack
• Session-Based Memory: Each candidate’s interview stored and managed separately
• Professional Summaries: AI-generated, downloadable assessments
• Robust Data Handling: Secure, session-based storage, easy clearing

🛠️ Tech Stack

• Python 3.10+
• Streamlit (chat UI)
• FastAPI (backend API)
• Groq/Llama3 (or OpenAI GPT API)
• Qdrant (vector DB for session state)
• PyMuPDF (resume PDF parsing)

🔧 Setup & Installation

1. Clone repo:
   git clone https://github.com/korupolujayanth2004/talentscout-ai.git

2. Environment:
   python3 -m venv venv && source venv/bin/activate

3. Install requirements:
   pip install -r requirements.txt

4. Run Backend API:
   cd backend
uvicorn main:app --reload

5. Run Frontend/Chat:
   cd frontend
streamlit run app.py

❓ Usage

• Upload your resume or enter info manually
• Chat with the AI for a fully interactive “real interview” experience
• Download summary and feedback afterwards

🎯 Prompt Engineering Notes

• Information extraction prompt guides the LLM to only ask for missing fields
• Tech question prompt takes in tech stack, requests only conceptual/practical questions
• Summarizer prompt polishes the conversation history and returns a clear HR report

🏆 Performance and Extensions

• API responses are cached for common questions (performance)
• Can be extended with emotion/sentiment analysis and multilingual support

📄 License

MIT