Period

2025 — Present

Role

Operations Intern → Consultant

Scope

Ops · AI · Build

Status

In delivery

From procurement intern to building regulated QC software for the same company.

Started by rewriting Saol's procurement workflow as an operations intern. Two summers later, building Kitscan — a barcode-driven quality control scanner for swab/brush kit manufacture, with the audit-trail discipline a regulated environment demands.

Active Project — 2026

+15%

Procurement Productivity

190

Kits Verified Per Run

PASS/FAIL Decision Time

01The problem

Manual QC was the bottleneck. And the risk.

Saol Therapeutics builds clinical kits — physical assemblies of swabs, brushes, labels and packaging that ship to research and treatment sites. Each kit carries a unique configuration of patient and sample IDs that must match an authoritative kit-construction sheet before release. A single mislabel can compromise a study.

The pre-existing process was paper-driven: an operator visually compared scanned barcodes against a spreadsheet, line by line, kit by kit. Slow on a good day. Brittle on a bad one. And critically, the audit trail of who-verified-what-when lived in handwritten records — not what a regulator wants to see.

"In a regulated environment, the audit trail isn't documentation — it's the product. If you can't show how a kit got verified, you didn't verify it."

02Approach

Build the scanner the QA team can actually defend.

The spec wasn't elaborate. It was specific. Barcode in, expected-set lookup against the authoritative kit list, binary PASS/FAIL within two seconds, immutable log entry per event. Every decision in the design serves either the operator (fast, unambiguous feedback) or the auditor (a record they cannot doubt).

A / 01

Earn the brief the slow way

Started two summers ago as a procurement intern — sat with finance and IT, rewrote sourcing workflows, shipped a +15% productivity gain. That earned the trust to be handed the QC problem when it surfaced.

A / 02

Spec to ten functional requirements

Captured the scope into ten binary, testable requirements — barcode capture, kit identification, expected-set lookup, PASS/FAIL logic, operator feedback, persistent logging, re-scan handling, source-of-truth writeback, error reporting, and audit trail integrity. Each one signed off by QA before any code shipped.

A / 03

Design for the auditor, not the demo

Every verification event logged with Kit Number, scanned IDs, outcome, timestamp, named operator. PASS never overwrites a prior FAIL. The log is exportable on demand. Boring features, but they're what makes the system defensible.

A / 04

Two-second feedback loop

An operator scanning a kit gets a clear visual (and audible) PASS or FAIL within two seconds of the final scan. No ambiguous states, no spinning loaders, no second-guessing. The system either passed it or it didn't.

Built with

Next.jsSupabase TypeScriptTailwind Barcode SDKAudit-Trail Schema

03Outcomes

From two summers of trust to a live build.

Phase 1 win

+15%

Procurement throughput gain across Finance & IT — the work that earned the bigger brief.

Kit coverage

190

Distinct kit configurations the scanner verifies, sourced from the authoritative kit-construction sheet.

Operator decision time

Maximum time between final scan and PASS/FAIL verdict — fast enough to keep packing-line throughput.

Kitscan is in active delivery as of mid-2026, on a 60-hour build estimate with monthly hours-actual billing. The maintenance arc — patches, content updates, infrastructure — is contracted separately under a monthly retainer.

04What I'd take elsewhere

The trust ladder is real. Climb it on purpose.

I didn't walk into Saol and pitch a QMS tool — that wouldn't have worked. I started with a smaller, cheaper problem (procurement), shipped it well, and stayed close to the people whose problems were bigger. By the time Kitscan came up, the answer to "who builds this?" wasn't a search.

If you're trying to win regulated or risk-averse work, this is the path. Make the small thing right. Stay around. The big briefs find their own way to people they already trust.

§ Next case study

Ubisense — shipping logistics software as an intern

→