Where a model has to answer to an experiment.
Glint Micro works on problems where the maths gets physical and the error bars are part of the design. Small studio, long horizons, tools we use ourselves.
Knowing what a number means
Most of the work in an experiment is deciding what counts as a reading. We build the parts that keep that decision explicit instead of buried in a pipeline.
Models that stay falsifiable
A simulation is only useful if it can be wrong in a way you would notice. We spend our time on the seam between a learned model and the physics it is standing in for.
Software with a grain to it
Everything we release started as something we needed on a Tuesday. Local-first, legible file formats, no account required to open your own work.
A workspace for people who keep their notes for years. Plain files on your own disk, a fast outline, and no cloud in the critical path.
- Files
- Plain .md in a folder you chose
- Sync
- Whatever you already use. None required.
- Platforms
- Windows
If your problem has instruments in it, write to us.
We take on a small number of collaborations a year, usually with groups who already have data and a disagreement about what it means.