Halo Collar
Role
Founding / Lead Designer — Softeq Development
Period
2018 — 2022
Metrics
Onboarding cut from 15 minutes to under 2
Domain / Status
IoT / Consumer
Shipped

Overview
Halo Collar is a GPS dog-safety wearable. As Founding / Lead Designer I designed the 0→1 mobile app, working directly with hardware and ML engineering to translate device constraints into product decisions.
The Ask
Design a consumer mobile app from zero for a hardware product whose core constraints — GPS/satellite connectivity, SIM-based firmware updates, virtual geo-fences — are set by physics and hardware, not by design.
Constraints
GPS/satellite connectivity: signal loss, delayed fixes, battery trade-offs
SIM-based updates: firmware changes ship slowly, UI has to tolerate stale state
Virtual fences / geo-fences as a core interaction, not a feature
New product category: no existing mental model to borrow from
Onboarding had to work for non-technical pet owners
What I Did — UX Architecture
Every hardware constraint became an explicit design decision instead of an apology in a tooltip. Connectivity gaps were designed as legible states ("last known location, 4 minutes ago") instead of hidden failures. Geo-fence creation was built as a direct-manipulation map interaction, not a settings form.
Interaction Design & Onboarding
Onboarding was rebuilt from 15 minutes to under 2 through iterative, research-driven usability testing — cutting every step that didn't directly get a fence set up and a collar paired. I scaled and led the growing design team through this period.
Key Screens
Live location with connectivity-state legibility
Direct-manipulation geo-fence creation on the map
Streamlined pairing & onboarding flow (15 min → under 2 min)
Alert and notification design for fence breaches
Outcome
The company went on to raise $40M+ at a $400M valuation. The app holds 4.1★ (8.5K ratings) with 100K+ Google Play downloads. (Company outcomes shown for context — not claimed as a direct result of design work alone.)
What I'd Do Differently
Instrument connectivity-state edge cases earlier with real hardware in the field, rather than relying on simulated GPS dropout in early usability rounds.