Python Engineer

Stepio takes one screen recording and returns a polished demo video plus a step-by-step guide, in 47 languages. The pipeline behind that is Python: it cuts clips, generates voiceover, renders captions, mixes audio and muxes the final file. You would own it.

What you will do

• Build and maintain the rendering pipeline that runs on Modal, using FFmpeg for cutting, muxing and encoding.
• Work on text-to-speech across multiple providers, including word-level timing so audio stays in sync with the picture.
• Improve caption rendering across writing systems, from Latin to Tamil and Sinhala, where font selection and text direction both matter.
• Keep render times predictable. A change that adds thirty seconds is a change every user feels.
• Handle the unglamorous parts well: retries, partial failures, and jobs that die halfway through a large file.
• Instrument the pipeline so that when a render looks wrong, we can tell which stage did it.
• Review output, not only code. Watching the video you produced is part of the job.

What we look for

• Strong Python, and comfort reading a stack trace from a subprocess you did not write.
• Real experience with FFmpeg or media processing. You should know why a re-encode is expensive.
• Care about correctness in systems whose output is a file nobody reviews line by line.
• Willingness to debug by inspecting artefacts, not only logs.
• Judgement about when to optimise and when the slow version is fine.

Nice to have

• Modal, or other serverless container platforms.
• Audio work such as loudness normalisation or forced alignment.
• Supabase or Postgres beyond basic CRUD.
• Having shipped something where latency was a product requirement.