Four years of support tickets taught me what to build.

I'm a new grad heading into product management, by way of IT. I hear what's broken, then I scope the fix, ship it and measure it.

Davis Oduor

Davis OduorB.S. Computer Science, SF State, May 2027

Decisions I’d make again

Before product, I spent four years in IT support, which is where I learned that most bad software is a decision nobody wrote down. These are mine, written down: what I chose, what I passed on, and why.

Thuolo

Read employers' own career boards instead of job aggregators

Chose
Crawl Greenhouse, Lever, Ashby and Workday directly, every 4 hours.
Over
Aggregator feeds, which are faster to build and already cover every company.
Because
Aggregators are where the closed and reposted listings come from. Now 98% of live postings come straight from the employer.
Thuolo

Measure what the crawler drops before adding cities

Chose
A script that counted every posting we'd lose, by state.
Over
Adding the biggest metros next.
Because
It found 11,785 lost US roles, and that, not city size, set the order. We went from 10 metros to 26.
Jarito

Use the calendar feed, not the Canvas API

Chose
The read-only calendar link every student already has.
Over
The API, which gives richer data.
Because
Most schools don't let students create API tokens. The better data wasn't worth losing most of the users.
Desk Check

Start with capture-once, not the whole audit

Chose
A tool that records each finding once and generates its ServiceNow ticket.
Over
Automating the entire audit end to end.
Because
It was phase one of a tiered automation roadmap: the retyping was where the four minutes went, so that came first.

What I’ve built

Each one started as somebody’s problem.