AIcruiter
AI-powered interview agent to automate recruitment, reduce bias, and save time.
Created at GenAI Genesis 2025 Hackathon
๐ Inspiration
Hiring is one of the most critical yet flawed processes in organizations. Human interviews often introduce unconscious biases, and HR teams spend countless hours manually scheduling and conducting interviews. We wanted to rethink this โ what if interviews could be efficient, unbiased, and available 24/7?
Introducing: AIcruiter โ a hyper-realistic AI-powered interview agent that automates the entire screening and interviewing pipeline while ensuring fairness, consistency, and customization for each company.
๐ What it does
AIcruiter is a fully autonomous AI agent that conducts live, interactive job interviews via video call. Here's how it works:
- HR Customization Portal โ The HR team logs into our webapp to customize the AI agent with their company's values, job descriptions, and tone.
- Automated Scheduling โ Once configured, AIcruiter sends candidates an email with a WebRTC video call link.
- Real-Time Interview โ Candidates can join at their convenience. AIcruiter engages them in an interactive, human-like interview, assessing both technical and behavioral competencies.
- Streaming Intelligence โ Leveraging Pipecat, our system streams real-time speech-to-text, LLM-generated responses, and text-to-speech for ultra-low latency conversation.
- Scoring & Insights (Coming Soon) โ Interviews are analyzed using a semantic scoring model, and results are made available to HR for data-driven decisions.
๐ ๏ธ Tech Stack
| Domain | Tools |
|---|---|
| Frontend | HTML/CSS/JS, Bootstrap 5, Jinja2 |
| Backend | Flask (Python) |
| AI & ML | Google Gemini 2.0 (Flash), RAG |
| Speech | Deepgram (STT), Cartesia (TTS) |
| Media Pipeline | Pipecat (Real-time Streaming), Tavus (Video), Daily (Transport) |
| Resume Parsing | PyPDF2 |
| Other Libraries | aiohttp, Loguru, Pytorch |
๐ก Key Features
- ๐๏ธ HR Control Center to customize AI agent personality, values, and questions
- ๐ง Conversational AI Agent that conducts live interviews with real-time STT โ LLM โ TTS pipeline
- ๐ง Email Automation to send interview invites with secure video links
- ๐ฅ WebRTC-based Interviews using Daily + Pipecat for minimal latency
- ๐ Scoring Engine Direct Integration (Coming Soon): Semantic model to rate candidates based on responses
๐ฆ How we built it
We stitched together multiple components into one seamless real-time pipeline:
- Used Pipecat to create a low-latency streaming pipeline from STT โ LLM โ TTS
- Used Gemini Flash (Google Generative AI) to simulate a smart, responsive interviewer
- Integrated Deepgram for accurate, streaming speech-to-text transcription
- Used Cartesia for lifelike voice synthesis (TTS)
- Managed video calls with Daily and video service with Tavus
- Built the frontend HR portal and candidate experience using Flask + Bootstrap
- Set up secure interview request flows with email integration and link-based scheduling
๐ Customization & Privacy
Every company has unique needs โ AIcruiter allows full custom agent configuration (tone, values, job role). All sessions are private and encrypted via Daily's secure WebRTC infrastructure.
๐งช What's next?
- โ Semantic Interview Scoring Model integration into the web dashboard
- ๐ Feedback Loop for AI improvement based on HR feedback
- ๐ OAuth & Company-Level Access Management
- ๐ Multilingual Support for global hiring

Log in or sign up for Devpost to join the conversation.