Engineering evidence

Scoped, dated, and methodologically qualified — not just numbers

A number on its own is a claim. The same number with its scope, its date boundary, and its method attached is evidence. Everything below is the second kind: what each figure covers, over what period, measured how — and, where it matters, what that measurement cannot show. Every row links to the case study that explains it in full.

The record

StatisticValue
Years in software engineering4+Software-developer and engineering roles since January 2022; the earlier IT-analyst and field period (Mar 2021 – Jan 2022, then a software-developer year at the same company) is stated separately, not added in — Counted from the first software-developer title on the CV (Jan 2022) to the present; rounded down, never up
Design specifications written255Cendra, tenure-scoped (Aug 2025 – Jul 2026) — Self-counted in private company repositories; consistent with the CV
Implementation plans written207Cendra, tenure-scoped (Aug 2025 – Jul 2026) — Self-counted in private company repositories; consistent with the CV
Production runbooks authored17Cendra, tenure-scoped (Aug 2025 – Jul 2026) — Self-counted; each one governs a coordinated release
Production releases coordinated377Cendra, across four repositories, tenure-scoped — Self-counted; roughly one release every 22 hours across the tenure
Languages shipped, including Arabic RTL13Cendra product surface, tenure-scoped — Self-reported product scope
LILA commits since founding~9,500LILA, solo project (Feb 2026 – present) — Approximate, self-counted via git on the main branch (9,550 on 4 September 2026); unverified by anyone else — it's a private repository I own
Archive Engineering commits257 of 289Archive Engineering, solo-author project — Self-counted; 31 of the 289 were made by an AI coding tool under my direction — stated here, not netted out of the total
GitHub contributions6,562GitHub profile, Aug 2025 – Jul 2026 window — Directly measured from GitHub's own contribution graph; covers LILA, Archive Engineering and ReadyFly only — Cendra's Azure DevOps history is not represented. A dated measurement: the graph below refreshes at build time and will drift past this figure
Surviving frontend authorship, Cendra84.7%Cendra frontend, sampled at tenure's end — git blame -w -M against origin/prod, a deterministic 1-in-7 file sample
Tenure-scoped frontend commits, Cendra80.7%Cendra frontend, production branch, tenure-scoped — Commit-count share, not surviving-code share — a different measurement from the row above
Surviving backend authorship, Cendra23.2%Cendra backend, sampled — joined roughly six months into an eighteen-month-old codebase — Sampled git blame; a substantial contributor to that codebase, not its architect
Cendra's seed round$1MAnnounced 11 February 2026, led by Revo Capital with follow-on from Heartfelt Capital, Türkiye Development Fund and APY Ventures — Company-announced public record; the official announcement names me as founding engineer. Not a personal metric.
Unnecessary push notifications eliminated~19,000/dayCendra, production telemetry — Measured over a ten-minute observation window and extrapolated to a daily figure — not a directly observed 24-hour count
ReadyFly web application commits157 of 209ReadyFly web application (Jun 2025 – Jun 2026) — Self-counted, roughly three quarters; also a contributor to the API (31 of 209 commits), not its principal author

How to read these numbers

Commit count is not surviving authorship

A commit count measures participation — how many times someone touched a repository. It says nothing about whether that code is still there. Surviving authorship measures what actually ships today, attributed to whoever last substantively wrote each line. The two disagree whenever code is rewritten, reformatted, or replaced, which is most of the time in a codebase more than a few months old. Where I have both figures for the same repository — Cendra’s frontend, at 84.7% surviving versus 80.7% of tenure-scoped commits — I show both, because they answer different questions and neither one alone is the honest answer.

The blame method, and why it is conservative

Surviving-authorship figures come from git blame -w -M against the production branch, sampled rather than run against every file for tractability. -w ignores whitespace-only changes, so a reformat by another engineer does not silently reassign lines that were never really theirs. -M detects moved and copied lines, so a file split or a component moved to a new path does not read as someone else’s new authorship. Both flags make the figure harder to inflate, not easier — which is the direction a claim like this should err.

Why there is no “verify live” button

LILA and Archive Engineering are private because they are mine to open; Cendra and ReadyFly are private because they belong to companies that have not agreed to publish their source, their infrastructure, or their customers’ data, and I would not ask them to just to make a portfolio more convincing. Confidentiality here is not a gap in the evidence — it is a constraint the evidence is built to work within: every figure states its method precisely enough that you can judge what it can and cannot prove without needing the repository itself.

Why Cendra does not appear on GitHub

Cendra’s repositories were hosted on Azure DevOps, not GitHub, for the whole of my tenure there. The 6,562 contributions recorded in the table above cover August 2025 – July 2026 and are real and directly measured, but they span only the three projects I hosted on GitHub — LILA, Archive Engineering, and ReadyFly. Cendra’s commit history, releases, and specifications are real too; they are simply invisible to a tool that only reads GitHub, which is why this page states each figure’s source rather than letting one graph stand in for all of it.

GitHub activity

GitHub activity for the last two calendar years, as of September 2026; the data refreshes when the site is built. Cendra was hosted on Azure DevOps and is not represented here.

Capability × ownership

Capability × project
CapabilityLILAArchive EngineeringCendraReadyFly
Product/UXOwnerOwnerOwnerOwner
FrontendOwnerOwnerOwnerOwner
MobileOwnerOwnerOwner
BackendOwnerOwnerContributorContributor
AI layerOwnerOwnerContributor (bounded)Not owned
Release & opsOwnerOwnerOwner

“Owner” is sole or principal author of that capability on that project. “Contributor” means a real but partial share, stated precisely on each project’s case study. “Contributor (bounded)” marks Cendra’s AI layer specifically: I built the product surfaces and the transport around Cendra’s AI capabilities, not the underlying orchestration, which was owned by another engineer. “Not owned” marks ReadyFly’s semantic-matching and candidate-evaluation models, owned by its backend and ML engineers. An em dash marks a capability not exercised on that project.

This website’s codebase

StatisticValue
Lines of TypeScript powering this website7637