Back to projects
ProductivityCompleted

HotSeat AI

Privacy-first AI interview coach with voice practice, real-time body language analysis, and resume optimization: all running locally.

January 10, 2026View Code
TauriReactFastAPIPythonOllamawhisper.cppMediaPipe
Share:

The Problem

Job interviews are high-stakes situations where practice makes a real difference. But practicing alone lacks feedback, and coaching services are expensive. I wanted a way to practice with realistic AI interviewers who could evaluate my responses without uploading my career history to cloud services.

The Solution

HotSeatAI was envisioned as a desktop interview coach that runs 100% on your machine:

  • Voice Conversations: Speak naturally with AI interviewers
  • Real-Time Feedback: Body language and eye contact analysis via webcam
  • Resume Tools: ATS optimization and job-specific tailoring
  • Complete Privacy: No cloud, no telemetry, no data collection

Features

Interview Practice

20+ AI Personas across industries:

  • Tech interviewers (behavioral, technical, system design)
  • Business roles (consulting, finance, marketing)
  • Various seniority levels and interview styles

Practice Modes:

  • Quick Start: Jump into random practice
  • Targeted: Paste a job description for tailored questions
  • Random: Surprise questions to test adaptability

HotSeat AI Interview Dashboard

Performance Analytics

After each session, receive detailed scoring:

  • Clarity: How well you communicated
  • Confidence: Vocal patterns and body language
  • Technical Accuracy: For role-specific questions
  • STAR Method: Structure of behavioral answers

Body Language Feedback

Real-time webcam analysis using MediaPipe:

  • Posture assessment
  • Eye contact tracking
  • Nervous habits detection

Resume Optimization

  • ATS compatibility scoring with specific fixes
  • Keyword gap analysis against job descriptions
  • Professional critique and suggestions

Tech Stack

LayerTechnology
DesktopTauri (Rust wrapper)
FrontendReact 18, TypeScript, Zustand
BackendPython 3.11+, FastAPI
DatabaseSQLite with SQLAlchemy
LLMOllama (Gemma 3, Qwen 2.5)
Speech-to-Textwhisper.cpp
Text-to-SpeechPiper TTS
VisionMediaPipe

System Requirements

SpecMinimumRecommended
RAM8GB16GB+
Storage10GB20GB
GPUOptionalImproves LLM speed

Why Local?

Interview preparation involves sensitive information:

  • Your resume and career history
  • Salary expectations and job targets
  • Recorded practice sessions

By processing everything locally:

  • Your data never leaves your device
  • No usage limits or API costs
  • Works offline (great for travel)
  • No risk of cloud service discontinuation

Development

Built with spec-driven development over several weeks, including:

  • Full interview flow (greeting → Q&A → candidate questions → closing)
  • Session recording and playback
  • PDF export for session reviews
  • Progress tracking with trend analysis

Related Articles

FullStackGrower

Appearance

Connect