A working portfolio of instruments, terminals and experimental rigs we've engineered for pharmacies, labs and retail operators. Each project documents the constraint, the build, and the measurement.
From peptide cold-chain monitors to card-terminal firmware refactors. We publish what worked, what didn't, and the numbers behind each decision.
A four-channel thermal logger with a magnetometer-equipped lid sensor, designed to flag freeze-thaw excursions in research freezers. Sampling at 0.5 Hz over BLE, with onboard SD buffering for lab deployments without reliable Wi-Fi.
Read case studyA touchscreen point-of-sale for seasonal farm shops, designed to operate on intermittent LTE at remote markets. End-of-day reconciliation builds a single signed transaction bundle that uploads once connectivity returns, preserving audit continuity.
Read case studyAn analysis pipeline that ingests raw HPLC traces and returns fitted first-order degradation constants with bootstrap confidence intervals. Replaced a spreadsheet workflow that had drifted out of validation across two labs.
Read case studyA reconfigurable cash desk skeleton in milled aluminium, designed to accept swappable scanner, printer and payment modules. Built for a pharmacy chain rolling out 64 stores in a single quarter.
Read case studyA refactor of legacy payment terminal firmware for a Polish acquirer, splitting a 90 000-line C codebase into testable modules. Cut regression-test time from 11 hours to 38 minutes per release candidate.
Read case studyA reference kiosk build that any integrator can re-skin in a week. Documented the build openly, including the four weeks we spent on the wrong scanner mount, so others don't repeat our mistake.
Read case studyThree habits we bring into every engagement, regardless of stack. They're unglamorous and they are the reason the prototypes survive contact with a real environment.
Before any schematic or screen, we spend a week recording what the current system actually does — not what the brief says it does. Half of the problem is usually elsewhere.
We commit to a runnable, ugly, end-to-end version of the system by the third week. From there, every later decision is a refinement, not a leap of faith.
Every project ships with a written record of what broke, when, and how we caught it. The next team inherits the failure catalogue, not just the working code.
38
Hardware prototypes shipped since 2021
11
POS deployments in production retail
4
Published peptide stability studies