Media automation engineer. I build open-source tooling for multi-language YouTube publishing — dubbing, video assembly, and the API + browser pipelines that stitch it together.
Published on PyPI:
| Package | What it does |
|---|---|
| dubfit | Fit dubbed narration to a video scene by scene. Caps the speed change and carries the residual drift forward, so timing never slides and no scene ends up sounding rushed. |
| ytmla | Multi-language audio tracks for YouTube: the Data API half, plus the Studio automation the API doesn't expose. |
What I work on
- Dubbing pipelines — fitting translated narration to an existing cut without audible speed changes and without drift accumulating across a long film
- Localization at scale — titles, descriptions and audio tracks across languages, verified end to end
- The gaps in the API — YouTube has no endpoint for multi-language audio, so that half runs through the Studio UI; documenting what the API can and can't do is half the value
- Video assembly — scene-level rendering, captions, and multi-track muxing
Everything here came out of a real production run, not a toy problem. Where a component isn't proven, the README says so.
Python · ffmpeg · Google APIs · Playwright