AI-Powered Recruitment/Interview Platform

Project Purpose & Overview

What is it?
A modular, AI-powered recruitment/interview platform. The backend is built with FastAPI (Python), and the frontend uses Next.js (React/TypeScript). The system leverages advanced AI (OpenAI models, Whisper, TTS) to automate and enhance candidate interviews and assessments.

Who is it for?
Companies and recruiters seeking to streamline, automate, and improve the candidate interview process.

What problem does it solve?
It automates candidate interviews, provides AI-driven insights, and manages audio (speech-to-text, text-to-speech) interactions, saving time and improving the quality of hiring decisions.

What makes it unique?
- Deep AI integration (OpenAI, Whisper, TTS)
- Modular, microservice-ready backend
- Real-time audio processing and analytics
- Modern, responsive frontend

General Overview

Key Features: - Automated candidate interviews (audio/text) - AI-powered insights and analytics - Speech-to-text (STT) and text-to-speech (TTS) services - Usage tracking and reporting - Admin and user portals

Use Cases: - Screening candidates at scale - Generating interview transcripts and summaries - Providing instant feedback and analytics to recruiters

Tech Stack

Frontend:
Next.js 14, NextUI, Tailwind CSS, TypeScript, Framer Motion

Backend:
FastAPI (Python), Uvicorn, Poetry, Prisma ORM

Database:
MongoDB (via Prisma)

Infrastructure/DevOps:
Poetry for Python dependency management, environment variables for config, S3 for storage

AI/ML:
OpenAI (GPT-4, GPT-3.5, etc.), Whisper (speech-to-text), TTS (text-to-speech), integrated via API keys

How AI Powers the System

Role of AI:
- Converts candidate speech to text (Whisper) - Generates interview questions, summaries, and insights (GPT-4, GPT-3.5) - Converts text prompts to audio (TTS) - Tracks and analyzes usage for reporting

Models Used:
- OpenAI GPT-4, GPT-3.5 (NLP, summarization, question generation) - Whisper (speech-to-text) - TTS (text-to-speech, likely via OpenAI or another provider)

Integration:
- API-based (OpenAI keys in config) - Modular routers for STT, TTS, and interview insights

AI Tasks:
- Natural language understanding and generation - Audio transcription and synthesis - Analytics and reporting

Challenges & Solutions:
- Challenge: Real-time audio processing
Solution: FastAPI async endpoints, efficient use of OpenAI/Whisper APIs
- Challenge: Usage tracking
Solution: Dedicated Usage model in MongoDB, Prisma ORM

Technical Breakdown

Database:
- MongoDB, managed via Prisma ORM - Key model: Usage (tracks app, API, service, type, usage metrics, timestamps)

Backend Architecture:
- FastAPI app with modular routers for STT, TTS, and interview insights - Middleware for CORS and usage tracking - Prisma for async DB operations - S3 integration for audio storage

Frontend:
- Next.js app with modular components for user/admin portals - API integration for backend endpoints - Real-time updates and analytics

Key API Routes/Services:
- /stt (speech-to-text) - /tts (text-to-speech) - /interview-insights (AI-powered analytics) - Usage and admin endpoints

User Journey Walkthrough

  1. User visits the portal (Next.js frontend)
  2. User starts an interview (audio or text)
  3. Frontend records/interacts with user (audio capture, UI prompts)
  4. Frontend sends audio/text to backend (/stt or /interview-insights)
  5. Backend processes request:
  6. If audio:
    • Converts speech to text (Whisper)
    • Stores audio in S3
  7. If text:
    • Sends to OpenAI for question generation/insights
  8. AI generates response/insights
  9. Backend returns results to frontend
  10. Frontend displays results, analytics, or next steps
  11. Usage is tracked in the database

Text-Based Flowchart / Architecture

User
  ↓
Frontend (Next.js)
  ↓
[User Action: Start Interview / Submit Audio/Text]
  ↓
API Call → Backend (FastAPI)
  ↓                                  
[Route: /stt, /tts, /interview-insights]
  ↓
Middleware (CORS, Usage Tracking)
  ↓
Business Logic
  ↓
  ├─ If Audio: Whisper STT → S3 Storage
  ├─ If Text: OpenAI GPT (NLP, Summarization, Q&A)
  └─ TTS: Text-to-Speech Service
  ↓
Database (MongoDB via Prisma) ←→ Usage Logging
  ↓
Response to Frontend
  ↓
User Sees Results, Analytics, Next Steps

Conclusion

This project demonstrates a robust, modern approach to automating and enhancing the recruitment process with AI. By combining FastAPI, Next.js, and advanced AI models (OpenAI, Whisper, TTS), the team delivered a scalable, modular platform that streamlines candidate interviews and provides actionable insights for recruiters.

Key strengths: - Deep AI integration for real-time audio and text analysis - Clean, modular codebase for rapid iteration and scaling - Modern, user-friendly frontend

Future opportunities: - Expand AI capabilities (custom models, more analytics) - Integrate with more HR tools and platforms - Enhance real-time feedback and reporting