Device Integration Engineer

About usAuraEye.AI is building an AI co-pilot for eye examinations. Our software connects to the examination equipment an optical practice already owns, takes over the routine mechanics of subjective refraction and lens fitting, and keeps the optometrist in control of every clinical decision. We are a technical, Hong Kong–based startup. The clinical decision engine is device-neutral by design. Everything that touches a physical instrument sits behind a single driver contract — and that layer is what this role owns. The roleYou will own the boundary between our decision engine and the instruments in the exam room: the communication links, the device drivers, the capability model that tells the engine what each machine can and cannot do, and the on-machine validation that proves it all actually works. This is a hands-on, lab-based role. The instruments do not travel, and neither does the evidence. What you'll doBuild and maintain device drivers that let our engine operate ophthalmic instruments over RS-232 serial and TCP links, implementing a stable, vendor-neutral driver contract.Bring up new instrument models in our lab: establish and characterise the communication link, define the encoding and state model, and map it cleanly onto our driver interface.Build read-only data adapters that ingest results from autorefractors, lensmeters and corneal topographers across binary, XML and file-drop formats.Design and run structured experiments on real hardware: write down a falsifiable prediction first, then verify the physical outcome with an operator standing at the machine.Own the certification checklist for every driver you ship. A driver is not shippable because its tests are green — it is shippable when a person has confirmed the measuring head physically did what the driver claims.Build the harnesses, labelled datasets, golden-file regression tests and documentation that make each integration reproducible by someone who wasn't there.Work with our clinical lead so device behaviour matches the clinical rulebook, and with our CTO on the abstractions that keep new vendors cheap to add.What we're looking for5+ years in embedded, device, instrumentation or industrial software.Deep hands-on experience with serial (RS-232) and TCP device links — framing, encoding, checksums, timing, flow control, contention, error handling and recovery. You have debugged a link that "should work" and found out why it didn't.Strong Python (asyncio, pyserial) and a test-first instinct. You are comfortable owning a codebase, not just a script.Windows engineering competence: COM port enumeration and drift, device manager, UI automation (pywinauto / UI Automation), and running software reliably as a long-lived unattended service.Manual-driven discipline: you work from vendor documentation, cite section + page + revision, and you mark an unverified assumption as unverified instead of shipping it.Safety-relevant judgement: you understand why "the software reported success" is not the same claim as "the hardware moved," and you design your verification accordingly.Based in Hong Kong, or ready to relocate. This role lives next to the equipment.Nice to haveMedical device software experience — IEC 62304, design history files, or supporting regulatory submissions (HK MDCO / NMPA / TFDA).Ophthalmic, dental, laboratory or industrial instrument integration background.MQTT / AWS IoT Core, containerised deployment, OTA update systems.Experience designing a hardware abstraction layer that survived three or more vendors.Reading ability in Japanese or Korean for vendor documentation.