I build things that probably shouldn't exist yet.

Biology was mostly a wet-lab discipline, and code was a side tool.

Analysis pipelines and research tooling for sequencing data — turning raw reads into something a human can reason about.

The real bottleneck was never the biology — it was the data engineering around it.

The seam between code and living systems is where I want to keep working. This one isn't finished; it's a thread I keep pulling.

People fail at sales because they lack information.

An execution-based SDR trainer — reps run realistic drills and get scored, instead of watching another video course.

The product worked. Getting it in front of the right people was the harder problem.

Distribution is often harder than engineering. I now design for distribution from day one.

Chaos can be scored — even meme-coin chaos.

Non-custodial portfolio tracking with risk scoring. Read-only by design: your keys never move, the tool just judges.

A sharp little tool in a market that punishes tourists.

Constraint is a feature — "read-only, non-custodial" was the best product decision in it. Say what you won't do.