Cheetah is an AI-powered macOS app built to help software engineers practise for technical interviews. It listens to the conversation in real time and provides live coaching, hints, and suggested solutions while integrating with live coding platforms.
Cheetah combines local speech recognition with a frontier language model. Audio is transcribed on your own machine using whisper.cpp, Georgi Gerganov’s port of OpenAI’s Whisper, so transcription runs locally and quickly on Apple silicon. The transcript is then sent to GPT-4, which generates hints and full solutions to the interview problem being discussed. The app is written natively in Swift and SwiftUI, requires macOS 13.1 or later, and works best on a recent Mac with Apple silicon; capturing both sides of a call can be done with a free loopback audio driver such as BlackHole. Cheetah is open source under the permissive CC0-1.0 license, has over 4,200 stars on GitHub, and was last updated in 2025. You supply your own OpenAI API key.
Cheetah is free and open source under the CC0-1.0 license. The only running cost is GPT-4 usage, which OpenAI bills directly to the API key you provide.
Software engineers preparing for technical interviews who want realistic, AI-assisted practice sessions with real-time feedback on a Mac.