VideoBot

About this project

An automated pipeline that turns written stories into finished short videos. It generates narration and timed captions, layers background gameplay footage underneath, and assembles the result with FFmpeg. The engineering focus is orchestration: several AI models and services have to hand work to one another reliably, from text through speech synthesis and caption timing to final encoding, with a FastAPI service coordinating the stages. Getting captions and audio to line up consistently, and keeping the pipeline resilient when a stage fails, was most of the work.