The Agentic AI Voice Interview Platform is a production-ready system that conducts intelligent, voice-based AI interviews by analyzing job descriptions and candidate resumes. This platform combines a sophisticated multi-agent architecture with free, open-source speech-to-text (STT) and text-to-speech (TTS) models to create a seamless voice interview experience.
- Multi-Agent Orchestration: Powered by LangGraph, CrewAI, and Agno.
- Natural Voice Interaction: Uses Groq (Llama 3 70B) for intelligence and Edge-TTS for high-quality, free voice synthesis.
- Smart Knowledge Retrieval: Integration with ChromaDB for context-aware Interviewing.
- Real-time Performance: WebSocket-based communication for low-latency voice interaction.
- Python 3.11 (Required for compatibility with all AI and voice libraries)
- Node.js & npm (For the React frontend)
- Groq API Key — see
.env.examplefor setup - NVIDIA API Key — see
.env.examplefor setup
Please see setup.md for the detailed step-by-step setup guide.
-
Environment Setup:
cp .env.example .env cp .env.example backend/.env # Then edit both .env files with your actual API keys -
Backend:
python3.11 -m venv venv source venv/bin/activate pip install -r requirements.txt pip install reportlab # Run tests python test_all_features.py # Run server cd backend ../venv/bin/uvicorn app.main:app --reload --port 8000
-
Frontend:
cd frontend npm install npm run devThe frontend will be available at
http://localhost:5173.
Copy .env.example (at the project root) to .env in both the root and backend/ directories, then fill in your actual values:
cp .env.example .env
cp .env.example backend/.envThe .env.example file documents every variable, its purpose, and where to obtain API keys. A valid SECRET_KEY is required — generate one with:
python -c 'import secrets; print(secrets.token_hex(32))'Note:
SECRET_KEYmust be a high-entropy key. The application will refuse to start if it is missing or still set to a placeholder.
- Backend: FastAPI, LangChain, LangGraph, SQL Alchemy, ChromaDB.
- Frontend: React, Vite, TailwindCSS, Framer Motion.
- Voice: Edge-TTS (TTS), NVIDIA Riva/Deepgram (STT).