Case study 01 · AI + AR · Field quality
Giving field technicians an AI second pair of eyes
Evolved a paper-checklist workflow into a mobile platform where computer vision verifies equipment, ports, fibers and labels on the spot — before the crew leaves the site.
- Role
- UX Architect (and the platform's original iOS engineer)
- Timeline
- 2014 – 2026
- Platforms
- iOS · Android · Windows · Web
- Partners
- FAST engineering, QA, Global Services, field testers, external dev partners
Port 8 · label mismatch. Re-seat fiber and rescan.
- 12 yrs
- shaping one platform — from first line of code to UX architecture
- 4
- platforms unified: iOS, Android, Windows & web
- Real-time
- pass / fail feedback inside the technician's normal flow
Context
Installation and quality audits used to run on paper checklists and phone calls. Defects surfaced days later in an audit — often after the crew had moved on — triggering costly site revisits and rework.
The question
“How might we catch installation defects while the technician is still on site, without adding steps to an already demanding job?”
Approach
- 01
Digitize the job
Built the original FAST iOS apps that replaced paper checklists with guided installation and audit flows tied to live inventory records.
- 02
Scale across roles and platforms
Extended the suite to web, Windows and Android, standardizing flows for technicians, auditors, project managers and partners across regions.
- 03
Map the data before the screens
Mapped every ORA entity, status and API dependency so the vision model's output lands cleanly in the FAST record — work engineering leadership later called out as a key enabler of the app's UX.
- 04
Design for the hand, not the desk
Storyboarded onboarding, scan guidance, progress and pass/fail states for one-handed use in difficult conditions, then iterated with field testers through TestFlight builds.
- 05
Close the loop on the web — and beyond
Designed browser-based ORA review for quality primes and explored an XR version of the workflow with an external partner to test the art of the possible.
Key insight
The fastest fix is the one made before anyone drives away. Moving quality from a later audit into the moment of work changed the whole design brief.
Design decisions
Feedback in seconds, not days
Immediate pass/fail with a plain-language reason and a fix path replaces the deferred audit report.
Progressive disclosure for results
Summary first; port-level detail only appears when something fails — keeping cognitive load low on site.
The human stays in control
AI suggests, the technician confirms or overrides, and every decision stays auditable.
Outcome
ORA moved beyond proof of concept into an advanced field trial in early 2026, shifting quality checks from after-the-fact audits to in-process verification. Early field results point to earlier defect detection, fewer revisits and lower cognitive load. I presented the platform's decade-long evolution at Ciena's Tech Forum 2026.
Formally recognized twice: for “transformative” wireframes and guidance on the image-recognition program (2024), and for the ORA data mapping and CS+ journey redesign (2026).