Skip to contentSkip to contact
Dennis Khor
Recruiter mode

Live proofSampled on a schedule, failures shown

Live proof

Claims on this site are backed by numbers the server fetches on a schedule. Each reading below shows when it was sampled and where it came from, and failures stay visible.

Readings
19
Projects measured
7
Kinds of reading
6
Sampled
Hourly, Lighthouse daily

Every reading

19 readings: 5 healthy, 2 listed, 5 counted, 1 slow, 4 rate limited, 1 private, 1 not sampled yet.

This site

Commit rhythm

Commits per week from each project's git history, every row on the same weeks, from the week of to the week of . Ingested from source on .
  • GeoTag ProGeoTag Pro: 13 commits in 8 active weeks between 17 Nov 2025 and 12 Oct 2026. Busiest week: 3 commits, week of 22 Jun 2026.
  • FaceHushFaceHush: 133 commits in 10 active weeks between 17 Nov 2025 and 12 Oct 2026. Busiest week: 87 commits, week of 18 May 2026.
  • TeowkeeTeowkee: 45 commits in 7 active weeks between 17 Nov 2025 and 12 Oct 2026. Busiest week: 23 commits, week of 15 Jun 2026.
  • Rossi ItalianRossi Italian: 100 commits in 5 active weeks between 17 Nov 2025 and 12 Oct 2026. Busiest week: 42 commits, week of 22 Jun 2026.
  • denniskhor.comdenniskhor.com: 2 commits in 1 active week between 17 Nov 2025 and 12 Oct 2026. Busiest week: 2 commits, week of 12 Oct 2026.

How each reading is measured

Site status

The server that renders this site sends a GET request to each live site and follows redirects. The reading is the final HTTP status and the time until the response headers arrived.

A 2xx answer within 1500 ms is healthy, a slower one is marked slow, and anything else is failing. The request gives up after 8 seconds and says so.

Lighthouse

Google PageSpeed Insights runs Lighthouse on an emulated mid-range phone and returns four category scores out of 100: performance, accessibility, best practices and SEO.

Each reading is one run, not the best of several. Lab scores move a few points between runs, and any score under 90 is flagged. The request gives up after 45 seconds. When PageSpeed Insights refuses a run, the row says rate limited and waits for the next try within the hour, and it is not counted as the site failing.

App Store

A lookup by bundle id against Apple's public iTunes lookup API. When Apple returns no result, the reading says the app is not listed yet instead of hiding the row.

Google Play

A GET request for the public Google Play listing page. HTTP 200 means the listing is live and HTTP 404 means it is not listed yet. Only the status code is read.

GitHub

GitHub's public REST API gives the repository's last push time and the conclusion of its most recent workflow run. A passing run is healthy, a failed run is failing, and a cancelled or unfinished run is flagged. A repository with no workflow runs says so instead of claiming to be healthy.

Private repositories answer anonymous requests with 404, so they show as private rather than as missing.

Test counts

Test cases are counted statically in each project's source by the ingest script and committed with this site, together with the day they were counted.

These are not live: they change only when the site is rebuilt from fresh source. A project that has not been counted yet shows as not counted, never as zero.

Schedule

Status, store, GitHub and test-count readings are cached for an hour. The first visit after the hour gets the previous values while a fresh sample runs in the background. If nobody visits for a day, the next visit waits for a fresh sample instead of seeing day-old numbers.

Lighthouse runs are slow, so they are cached for a day and expire after a week. A failed run is retried within the hour, so a rate limit or a slow audit does not sit on the page for a day.

The cache is shared by every page and every server instance, so each schedule samples once rather than once per page. A scheduled job can also mark every reading stale through a signed request, so the next visit resamples everything.

The sampled time on each row is the real one, so a cached value is never passed off as a fresh one.

Honesty rules

  • Nothing here is typed in by hand. Every value comes from a request the server made or from a file generated from source.
  • A failing check is shown as failing, with the reason, until a later sample passes.
  • A reading that could not be taken says why: an offline build, a private repository and an app that is not listed yet each have their own state.
  • Every value shows when it was sampled and links to where you can check it yourself.
  • No number animates up to a target. What you see is what the last sample returned.

Ask about fit

Paste a job description. You get a verdict, the evidence that matches it with citations, and what the sources do not show.

Open to senior AI and software engineering roles: on site or hybrid in Sydney, or remote for Australian and US teams.

Email meBook 15 minCV