Back to projects

TK AI Interviewer

Real-time voice AI interview platform

A real-time, voice-based AI interview platform where candidates speak live with a Gemini-powered AI interviewer and receive an automatically generated evaluation report.

Full-duplex
live voice AI
16kHz/24kHz
capture / playback
0hr TTL
report cache
Live interview session screen showing the AI interviewer's voice activity, the candidate's video feed, and a live transcript panel
Live interview session
HR dashboard listing all interview sessions with total/pending/in-progress/completed stats
HR dashboard
Generated interview report with a question-by-question breakdown, ratings, and HR notes
Generated interview report
Next.jsReactTypeScriptTailwind CSSPythonFastAPIGoogle ADKGemini Live APIWebSocketJWTSQLiteCloudflare R2ResendDocker

Manual first-round interviews (HR screening calls) are time-consuming, hard to scale consistently, and create scheduling bottlenecks between recruiters and candidates. Companies need a way to conduct structured, consistent voice interviews at scale without a human interviewer present for every session.

Built a full-stack platform where HR creates a personalized interview session (role, company, custom questions) and sends the candidate a unique link. The candidate joins a live, bidirectional voice conversation with an AI interviewer powered by Google's Gemini Live API, complete with preflight device checks, reconnection handling, and optional session recording. After the interview, the system automatically generates a structured Markdown evaluation report and emails it to HR, with the full transcript and recording available in an HR dashboard.

01HR creates a session
02Candidate joins
03Live conversation
04Report generation
05Review