UL Solutions · Fremont, California · May – Aug 2026

Three days of Ctrl+F. Now about thirty minutes.

I came back to UL as a product management intern, the first point of contact for five labs. The slowest thing I saw was a data review done by hand. So I built the checker that does the cross-checking.

Product management internReturning, promoted5 labs≈ 600 reports a week

Find the three values that don’t match the master table.

Report · sample plot pages

Plot	Band	Position	Gap mm	SAR 1g

Master test tablesource of truth

Plot	Band	Position	Gap mm	SAR 1g

0:00reviewing

Click a row in the report to mark it as wrong.

Sample data, made up for this page. It is not the lab’s tool, its reports or its numbers. The real checker runs 13 checks on every plot page; the latest version, v21, runs 19.

What the summer looked like

Four parts, in the order they happened.

Handed to me

First point of contact for any problem across five labs, and the product half: building internal software that makes the labs faster.

Noticed

Before a report went out, someone cross-checked every value against the master test table with Ctrl+F. About three days a review, roughly 180 engineer-hours a month.

Built

A checker that reads each report against the master table, runs 13 checks per plot page (19 in the latest version) and flags what doesn’t match. Then a companion tool that matches positions as exact, likely or needs review.

Changed

≈ 30 min

down from about three days, across all five labs. People still do the review; they start from the flags.

The rest of the job

Less visible, just as much of the work.

Five labs, one contactThe single point of escalation for lab issues: setting the fix order each day, and twelve weekly briefings to lab leadership on progress and risks.
Product workInterviewed lab engineers and managers, wrote the requirements and acceptance criteria, piloted with three engineers and fixed the nine issues they found before rollout.
Daily toolsSQL against the lab’s test database, Python and Flask for the checker, and Claude for a lot of the typing.
Handing it overTracked adoption, around 600 reports a week, in a weekly summary for lab leadership, and wrote the SOPs so the next person can run it.

Honest limit: an internal tool, so the real one stays at the lab. Both times are approximate. Before this, I spent the summer of 2025 here running SAR tests.