Kacper Wikiel

I run a one-person machine-learning lab in Warsaw.

Most of what I build reaches into the physical world: computer vision that reads the welds running the length of a country's fuel pipelines, a Raspberry Pi that picks aircraft out of the sky and fixes their GPS position from a single camera, OCR that runs end to end on the laptop in front of you.

I work mostly by directing fleets of AI agents. That is how one person keeps this many real systems running at once. I came up through physics, so I would rather measure a thing than trust a demo of it. The four 3090s under my desk mean I can run the experiments myself instead of taking a vendor's word for it.

The rest of the time I build my own things: a benchmark site that thirty thousand people read every month, a Polish text-to-speech voice, a film recommender. Building is how I think.

Selected work

Computer vision that reads the welds running the length of a national fuel-pipeline network. Building it, I turned up a sensor bias the earlier analysis had walked straight past.
A Raspberry Pi that finds aircraft, follows them with a gimbal, and fixes their GPS position from one camera in real time. Most of the AI worth having is going to run on small hardware like this.
A document platform with attribute-based access control and a hash-chain audit trail, for work where you have to prove where every record came from.

Things I'm building

Hardparse — an OCR API that reads tables, formulas, charts, and seals.
CodeSOTA — a price-aware benchmark site for machine-learning models, where the numbers come from running the models myself.
Bozenka — an offline Polish text-to-speech voice.
TasteRay — film recommendations that come out of an actual conversation.

Experience

PwC ML Engineer, contract2025
National Pipeline Operator ML / CV Engineer2024–25
Procter & Gamble Software Engineer, Generative AI2022–24
Igoria Trade Python Developer2021–22