Year · 2024–presentBy Ayush Niroula
A comprehensive platform for conducting interactive AI-powered interviews with voice recording capabilities—architected for scale on FastAPI, Celery, and AWS S3 with strong tenant data segregation. The Next.js frontend is tuned for a smooth conversational flow; the backend runs asynchronous media pipelines (Redis-backed Celery, multi-queue workers) so interviews, feedback, and storage stay reliable under heavy load.
Highlights
- Designed to support 10,000+ concurrent interview sessions with real-time AI feedback and isolated tenant data paths
- High-quality Text-to-Speech (TTS) via ElevenLabs with asynchronous audio preloading for fluid sessions
- FastAPI + Celery + Redis architecture for durable media processing, transcription, and scoring pipelines
- OpenAI-driven question generation and AssemblyAI (and related ASR paths) integrated into end-to-end interview flows
Tech stack
Next.js 15
React 19
TypeScript
Tailwind CSS
Redux Toolkit
FastAPI
SQLAlchemy
PostgreSQL
Redis- Celery
AWS S3
OpenAI API- ElevenLabs API
- AssemblyAI
Docker