InterviewMentor AI

Real-time, multimodal AI mock interviews with structured feedback.

Introduction

InterviewMentor AI is an AI-powered interview simulation product that helps candidates practise technical, behavioral and system design interviews through a live, conversational AI interviewer.

Problem

Candidates rarely get realistic interview practice with immediate, specific feedback on both what they said and how they said it.

Solution

A browser-based interview session where an AI interviewer speaks with the candidate in real time, watches a shared code editor, and produces a structured post-interview report.

User journey

Dashboard — review past sessions and start a new one.
Setup wizard — choose the target role and interview type.
Live session — talk with the AI interviewer while coding in a shared editor.
Analysis — receive a structured report and track progress over time.

Main features

Real-time AI interviewer using low-latency Gemini Live audio streaming
Role-specific questions for Frontend, Backend, DevOps and Product Management
Live code editor the AI can see and give hints on
Post-interview report covering technical accuracy, communication, behavioral signals and non-verbal cues
Progress dashboard with performance charts

My contribution

Designed and built the product end-to-end.

Architecture overview

Client-side single-page application — no custom backend.
Live session opens a WebSocket connection through the Gemini Live API and streams 16 kHz PCM audio and JPEG video frames.
AI audio responses are played back through the Web Audio API.
After the session, the transcript and recording are sent to Gemini 2.5 Flash for a structured JSON analysis.

Challenges and learnings

Keeping audio streaming low-latency enough to feel conversational.
Turning a free-form conversation into a structured, comparable feedback report.

IndependentHackathon build

AI · Hackathon project

DocFlow AI

AI visa document assistant: extract, check requirements and pre-fill forms.

React 18
TypeScript
Vite
Tailwind CSS