Ship and explain
Small changes, shipped often, each with the reason it exists. We hold ourselves to the standard we ask of models.
Careers
Help make AI accountable.
We are a small, technical team building the explainability and monitoring layer for production machine learning. If you care about models being right for the right reasons, we would like to hear from you.
How we work
Ship it, explain it, show it running.
Write things down
Decisions, designs and trade-offs live in documents anyone on the team can read. Good writing is part of the job.
Test on public data
Every capability is exercised on a public example with committed outputs before it is described anywhere.
Own the whole stack
From the attribution maths to the API to the dashboard that draws a forest, the same people see a feature through.
Who we look for
People who have shipped models and wished they could see inside them.
- Machine-learning engineersYou have put models into production and been asked why one of them did something. You know SHAP's limits as well as its uses.
- Backend engineersYou care about APIs that are boring in the best way: typed errors, idempotent uploads, predictable limits, honest documentation.
- Front-end engineersYou can make a 110-million-parameter network legible on a laptop screen and know when a chart is lying.
- Clear writersWhatever your role, you can explain a hard idea in plain language. It is most of what explainability is.
Open roles
Nothing posted right now. Introduce yourself anyway.
We are not advertising specific positions at the moment. If the work above sounds like yours, tell us what you would build at Kaimei and how you would build it. We read every note.
Open application
Engineering, research or anything else you think we are missing
Introduce yourself
See what we are building first.
The best introduction to Kaimei is the product. Run the Titanic example, read the client, and come with opinions.