Mattis Megevand

A local-first data explorer that runs SQL in the browser and turns each step into a shareable, replayable pipeline

How I combined quantization, pruning, and low-rank approximation to compress neural networks for FPGAs

A small Astro integration for editing rendered Markdown and saving changes back to the source file

A small image classifier that predicts whether an image should be inverted in dark mode

Converting gradient-boosted trees to SQL so predictions run where the data lives