[01]LIVE(2025)

AlgoMentor

AI-powered mock interview platform simulating real FAANG-style DSA interviews.

FAANG-style adaptive DSA interviews with voice & scoring

[01]SYSTEM OVERVIEW

AlgoMentor is a production-ready AI mock interview platform designed to simulate real FAANG-style DSA interview experiences with adaptive questioning and automated evaluation.

The Engineering Challenge

Engineering candidates lack access to realistic, adaptive mock interviews that mirror the pressure, timing, and feedback quality of actual FAANG technical interviews.

Implemented Solution

Built a full-stack system using Next.js and FastAPI, integrating Groq Llama-3.3-70B for LLM inference and AWS Polly for neural text-to-speech to create a voice-interactive interview experience.

[02]SYSTEM ARCHITECTURE

Next.js frontend with real-time voice interaction and structured interview flow
FastAPI backend with Groq Llama-3.3-70B for adaptive question generation and evaluation
AWS Polly neural TTS for realistic interviewer voice synthesis
PostgreSQL for persistent interview history and candidate performance tracking
JWT + Google OAuth authentication with secure session management

[03]TECHNOLOGIES UTILIZED

Next.jsFastAPIPythonPostgreSQLGroq Llama-3.3-70BAWS PollyJWTGoogle OAuth

[04]KEY FINDINGS & TAKEAWAYS

Structured prompt engineering is critical for consistent LLM scoring across correctness and communication dimensions

Neural TTS latency requires careful buffering strategy to maintain natural conversation flow