Published on 9 min read
Journal Safari #7. DeepMind's AlphaGenome reads a megabase of DNA, predicts thousands of genomic tracks, and matches or beats the strongest competing model on 25 of 26 variant tests, mostly by a little. It calls the direction of a variant's effect far better than the size, which makes it a scanner rather than a verdict, and a decent preview of where AI in biology is right now.
Read more →
Published on 9 min read
I wanted a cheap AI model to pick the one biology preprint worth reading out of hundreds. With no hand-labelled answers to check against, I graded 20 cheap model setups on how closely they matched an expensive one, Claude Sonnet 5, for under $3. The best match came from outside Sonnet's family, and the setup I now run costs about $2 a year.
Read more →
Published on 18 min read
Journal Safari #4. Eleven ESM protein language models were supposed to be near-interchangeable. They have different blind spots that do not overlap, and training each one on the others' best guesses puts sequence-only at the top of the ClinVar board. The gain comes off about 200 proteins, which is the part I keep chewing on.
Read more →
Published on 17 min read
Journal Safari #3. Pan-human Azimuth puts 27 million cells from 23 tissues on one annotation tree, then classifies them with a model 70× smaller than the foundation model I read in July — and wins. The contribution is not the model. It is the labels.
Read more →
Published on 11 min read
Weeks after YouTube erased me, I finally reached a human at Google. Their official solution to getting my music back involves deleting my family group and hand-migrating 16 years of Google Drive. Here's why that's worse than it sounds.
Read more →
Published on 8 min read
Kicking off Journal Safari, my biweekly close-read of one recent paper. First up: CellVQ, a 500-million-parameter single-cell foundation model that squashes every cell into a discrete code — and the peer reviews that quietly agreed with my doubts.
Read more →
Published on 10 min read
One June morning, Google erased my 16-year-old account without warning and rejected my appeal in minutes. Here is the timeline, and why I think an algorithm decided I was a problem.
Read more →
Published on 9 min read
Evernote suspended their developer-token API in January 2026. I needed to back up every Claude Code conversation anyway. So I turned email-to-note into a sync engine.
Read more →
Published on 7 min read
Battling grocery monopolies, testing AI coding agents, and planting a mutant sweet potato with my three-year-old.
Read more →
Published on 6 min read
I built a security wrapper so AI coding agents can run wild in Docker containers without accidentally joining a botnet or deleting your life's work.
Read more →