Voice Interview Agent

An AI-powered phone interview agent built at a hackathon in under 24 hours. The agent calls candidates over the phone and conducts a live voice interview — asking questions, listening to responses, and guiding the conversation — entirely autonomously.

How It Works

1. A candidate's phone number and target role are submitted
2. Twilio places an outbound call to the candidate
3. The voice AI agent conducts the interview in real time — asking questions, processing spoken responses, and adapting follow-ups
4. The conversation is orchestrated end-to-end using a Langflow pipeline

Tech Stack

Layer	Technology
Voice & Telephony	Twilio Voice
AI Orchestration	Langflow
LLM	OpenAI
Backend	Python / FastAPI

Architecture

Candidate Phone
      │
      ▼
  Twilio Voice  ──►  Langflow Pipeline  ──►  OpenAI LLM
      │                    │
      │            (Question generation,
      │             response processing,
      └─────────────follow-up logic)

Running Locally

1. Clone the repo and set up a virtual environment
2. Install dependencies:

   pip install -r requirements.txt

3. Add your API keys to a .env file:

   TWILIO_ACCOUNT_SID=...
   TWILIO_AUTH_TOKEN=...
   OPENAI_API_KEY=...
   

4. Start the Langflow server and import the flow
5. Run the FastAPI backend:

   uvicorn main:app --reload

Context

Built as a hackathon submission for a challenge centered on voice AI agents and Langflow. Scope: under 24 hours, solo/small team.