I Built My Own AI Video Clipping Tool Because the Alternatives Were Too Expensive

This article explores how an engineer built Video Wizard, an open-source AI tool that converts long-form video into short-form clips with subtitles, smart face tracking, and caption templates. Using a microservices architecture built with Next.js, Python, Whisper, GPT, MediaPipe, and Remotion, the project demonstrates how AI can automate transcription, detect viral moments, reframe video for vertical formats, and render polished clips for TikTok or YouTube Shorts. The post breaks down the technical architecture, key engineering challenges such as subtitle synchronization, and lessons learned building a production-grade video processing pipeline with AI as a coding partner.