Cheetah is an AI-powered macOS application designed to assist software engineers in practicing for technical interviews. It offers real-time audio transcription, AI-generated hints and solutions, and live coding analysis, targeting users preparing for interviews.

How It Works

The app utilizes OpenAI's GPT-4 for generating responses and hints, and Georgi Gerganov's whisper.cpp for local, real-time audio transcription via Whisper. This approach allows for on-device processing of audio, reducing latency and reliance on cloud services for transcription, while leveraging powerful LLMs for interview assistance.

Quick Start & Requirements

• Install: Build from source. Requires checking out whisper.cpp in ../whisper.cpp and installing SDL2 (brew install sdl2).
• Prerequisites: macOS 13.1 or later, Apple silicon Mac recommended for optimal performance.
• Audio Setup: Requires configuring a Multi-Output Device with BlackHole for capturing both sides of a conversation in video chat apps.
• Browser Extension: Firefox extension available for live coding analysis.
• API Key: Requires a personal OpenAI API key.
• Docs: Quick demo video

Highlighted Details

• Real-time audio transcription using local Whisper.cpp.
• GPT-4 powered "Answer", "Refine", and "Analyze" features.
• Browser extension for live coding platform integration.
• Supports refining generated answers based on highlighted text.

Maintenance & Community

No specific community links or maintenance details are provided in the README.

Licensing & Compatibility

The README does not specify a license. Compatibility is limited to macOS.

Limitations & Caveats

The app is intended for mock interviews only and may produce incorrect solutions. Debug mode significantly degrades transcription performance. The browser extension currently only supports Firefox.