Gig Bazar AI Interviewer

AI Interviewer who can screen candidates

Comment

Inspiration

Traditional hiring processes are time-consuming, expensive, and often inconsistent. HR teams spend countless hours conducting initial screening interviews, while candidates struggle to find convenient interview times across time zones. We were inspired to democratize access to quality job screening by creating an AI-powered interview platform that:

  • Scales instantly to screen hundreds of candidates simultaneously
  • Provides consistent, unbiased initial assessments
  • Works 24/7 across all time zones
  • Frees HR teams to focus on final-stage interviews and relationship building

We envisioned a marketplace where gig workers, freelancers, and job seekers could get instant, professional interview experiences, while employers could efficiently screen talent at scale.

## What it does

Gig Bazar AI Interviewer is a voice-based AI interview platform with two sides:

For HR/Employers:

  • Create detailed job postings with required skills, experience level, and interview duration
  • Automatically generate a dedicated AI interviewer agent for each position
  • Customize AI instructions (e.g., "focus on system design" or "assess leadership skills")
  • Share interview links via email, social media, or job boards
  • Review full interview transcripts with candidate details
  • Track all interviews from a centralized dashboard

For Candidates:

  • Click a shareable link - no account needed
  • See complete job details before starting
  • Have natural voice conversations with an AI interviewer tailored to the specific role
  • Experience structured interviews covering technical skills, behavioral questions, and Q&A
  • Complete interviews on their own schedule, anywhere in the world

The AI adapts every interview based on:

  • Job requirements and skills
  • Seniority level (junior/mid/senior)
  • Custom HR instructions
  • Candidate responses in real-time

## How we built it

### Frontend (React + TypeScript):

  • Built responsive UI with React and TypeScript for type safety
  • Integrated shadcn/ui components for a polished, professional interface
  • Implemented routing for HR dashboard, job creation, and public interview pages
  • Used Tailwind CSS for rapid, consistent styling
  • Connected to ElevenLabs Web SDK for real-time voice interactions

### Backend (Python + FastAPI):

  • Created RESTful API with FastAPI for high performance
  • Built modular services: job_service.py, interview_service.py, elevenlabs_service.py
  • Implemented dynamic prompt generation that adapts to each job's requirements
  • Automated ElevenLabs agent creation using their Conversational AI API
  • Integrated Firebase Authentication and Firestore database

AI Integration:

  • Integrated ElevenLabs Conversational AI for natural voice interactions
  • Designed comprehensive interview prompts with structured flow (opening → technical → behavioral → Q&A → closing)
  • Created automatic agent provisioning - each job gets its own dedicated AI interviewer
  • Built dynamic prompt system incorporating job details, skills, difficulty, and custom instructions

## Infrastructure:

  • Deployed on Google Cloud Platform
  • Used Firestore for real-time data synchronization
  • Implemented shareable token system for public interview access
  • Structured database schema for scalable multi-tenant architecture

## Challenges we ran into

  1. Dynamic Prompt Engineering: Creating a single prompt template that works for junior developers, senior architects, and everything in between required extensive iteration. We solved this by building a modular prompt system that adapts based on role requirements.
  2. Real-time Voice Latency: Initial implementations had noticeable delays in AI responses. We optimized by using ElevenLabs' low-latency conversational agents and implementing proper audio streaming.
  3. Agent Lifecycle Management: Deciding when to create, update, and delete AI agents for jobs. We implemented automatic creation on job posting and added manual management endpoints for flexibility.
  4. Transcript Storage & Retrieval: Handling potentially long interview transcripts efficiently in Firestore required careful schema design to balance query performance and cost.
  5. Public vs. Authenticated Routes: Balancing security (HR dashboard) with accessibility (public interview links) required careful authentication flow design.
  6. SDK Version Compatibility: ElevenLabs SDK updates required graceful error handling to ensure the system continues working even with API changes.

## Accomplishments that we're proud of

✅ End-to-End Working MVP: Built a complete, functional platform in hackathon timeframe ✅ Automatic AI Agent Creation: Seamless per-job agent provisioning with zero manual intervention ✅ Beautiful, Intuitive UI: Professional interface using shadcn/ui that rivals commercial products ✅ Natural Voice Conversations: Achieved smooth, low-latency AI interviews that feel human ✅ Scalable Architecture: Designed for thousands of concurrent interviews ✅ Zero Candidate Friction: No account creation - just click and interview ✅ Full Visibility: HR sees complete transcripts with timestamps and candidate information ✅ Extensible Design: Built modular services ready for marketplace features

## What we learned

## Technical Learnings:

  • Real-time voice AI requires careful latency optimization and error handling
  • Dynamic prompt engineering is an art - small changes dramatically affect interview quality
  • Firebase + React is incredibly powerful for rapid MVP development
  • TypeScript prevents countless bugs in complex React applications
  • Modular service architecture makes features easy to add and test

## Product Learnings:

  • HR users need customization but not overwhelming options
  • Candidates prefer no-friction experiences (no signup) over personalized accounts
  • Transcript visibility is more valuable to HR than automated scoring in early stages
  • Shareable links are the killer feature for distribution

## AI Learnings:

  • Conversational AI works best with clear structure (opening, technical, closing)
  • Custom instructions should guide, not override, base interview best practices
  • Voice-based screening reveals communication skills that text-based systems miss

## What's next for Gig Bazar AI Interviewer

Immediate Enhancements:

  1. AI-Powered Candidate Scoring: Automatic evaluation based on technical accuracy, communication clarity, and role fit
  2. Email Notifications: Auto-send interview invites and transcripts
  3. Interview Analytics: Dashboard showing candidate performance metrics, time-to-complete, and drop-off rates
  4. Candidate Accounts: Let candidates track their interview history and results

Marketplace Expansion:

  1. Gig/Freelance Matching: Extend beyond interviews to full job/gig marketplace
  2. Multi-Organization Support: White-label solution for recruitment agencies
  3. Video Support: Add visual component for roles requiring presentation skills
  4. Interview Scheduling: Calendar integration for hybrid AI + human interview workflows
  5. Payment Integration: Monetize through per-interview or subscription models

Advanced AI Features:

  1. Personality Assessment: Integrate psychometric evaluation
  2. Code Challenge Integration: Live coding assessments during voice interviews
  3. Multi-Language Support: Conduct interviews in candidate's preferred language
  4. Interview Templates: Pre-built agents for common roles (SWE, PM, Designer)

Enterprise Features:

  1. HRIS Integration: Connect with Workday, Greenhouse, Lever
  2. Compliance & Privacy: GDPR, SOC2, candidate data management
  3. Collaboration Tools: Multi-reviewer workflows, shared notes, hiring committees

Our vision: Become the default pre-screening layer for every job posting worldwide - making hiring faster, fairer, and more accessible for everyone.

Built With

Share this project:

Updates