I Built a Desktop AI Mock Interviewer with Speech, Code Editor & PDF Reports

#python #electron #ai #career

What I Built

A fully offline-capable Windows desktop app that simulates a real guided interview
session — from topic selection to a generated PDF performance report at the end.

You pick your topic (Programming, CS Fundamentals, SQL, Behavioral, etc.), set
the difficulty, and the AI interviewer starts asking questions. You can answer
verbally or type code in the built-in editor. When the session ends, you get a
structured PDF with your scores, feedback, and recommended next steps.

Tech Stack

Layer	Tech
UI	Electron
AI Engine	Python
Communication	IPC (stdin/stdout)
Speech-to-Text	Groq Whisper
LLM	LLaMA 3 via Groq API
PDF Generation	ReportLab
Packaging	PyInstaller + electron-builder

Features I'm Proud Of

Live speech transcription — answer questions out loud, Whisper handles the rest
Adaptive follow-up questions — the AI reads your answer and digs deeper
Built-in code editor — for programming/DSA rounds
Behavioural monitoring — tracks focus loss, face presence via camera, and filler words
PDF report — per-question scores, structured feedback, next steps

Why Desktop + Local-First?

Most AI interview tools are web apps that store your data on their servers.
I wanted something that runs entirely on your machine. You bring your own
Groq API key — no accounts, no backend, no data leaving your PC.

Architecture: Electron + Python over IPC

The trickiest part was getting Electron and Python to talk reliably.
I went with stdin/stdout IPC — Electron spawns the Python process and
they communicate via structured messages. It's simple, cross-process,
and works well with PyInstaller packaging.