Ten years across data science and product, three shipped products of my own, and the studies that decided what stayed in them. Working copy — this line is yours to rewrite.
Products
Each one exists because something kept breaking in the ordinary way of doing it. Each carries a pre-registered study of a question that mattered to its design.
A news pipeline that assigns each extracted claim a verification tier, keeping the strength of the evidence separate from how well the claim was reported, so a well-reported claim cannot be upgraded to a true one.
23.9%[17.3, 30.5] · 95% interval
of week-one Contested verdicts held up against a later verdict built from cited evidence, so about 3 in 4 of the claims marked Contested were true.
Local discovery for people who are past the touristy layer, built on a layered retrieval engine that reads intent from a plain question.
69.4%[52.8, 84.5] · 95% interval
of good answers lost when query interpretation is removed, the largest cost of any single layer, measured on an arm run too few times to pass the stability gate.
Synthetic user populations that browse a real web app and turn what they do into findings: rage clicks, confusion, pages nobody can find.
64/64qualifying sessions caught
Every session that produced the behavior its rage-click signal looks for was caught, and none was caught without it, in a study that maps where the instrument goes blind.
Studies
Designed before the data existed, run against my own products, and ending in a recorded decision. Every number cites the committed code or data it comes from.
VerifiedNews · decided 2026-10-09
Week-one Contested verdicts were re-judged against later cited evidence, and about 3 in 4 of those claims turned out true. The fix to the single-denial rule is decided and queued behind work that has a consumer for the tiers.
Sidestreet · decided 2026-09-28
A 9-arm ablation priced every layer of the engine. The strict filter stayed off: restoring it costs 16.7% of good answers to prevent a quality loss the study could not find.
Understudy · decided 2026-10-11
Five seeded defects, controlled doses, resampled populations. The answer draws a boundary: element-level detections earn trust, page-level signals and silence do not.
github.com/minihm
Every study lives in its own repository, with its frozen design, its data, its analysis, and a script that recomputes published numbers from source. The repositories open with each release.
Frozen design, 600-claim sample, re-scoring pipeline and verifier.
9 arms, stability gates, and the cost of every retrieval layer.
Seeded defects, synthetic populations, detection boundaries.
This site — content as typed collections, gates that fail bad builds.
$ python3 specs/verify_published.py → 30 checks OK, exit 0
About
I've spent ten years where data science meets product: enterprise data at Autodesk, then six years at 6sense working with enterprise accounts, where a value-measurement framework I built outside my job description became a product adopted across teams. Now I'm the sole data scientist at an early-stage AI company, building the data function from nothing and sitting in the product decisions it feeds.
The products on this page are mine — nights-and-weekends builds that turned into real systems. The studies exist because I wanted my own decisions held to the standard I'd demand at work.
Ask this site anything — a retrieval engine over everything here is in design. For now, the studies answer the hard questions.