Case 01 · Amazon Robotics
ToteASRS is Amazon Robotics' automated inventory storage and retrieval system. The robots handle retrieval — but humans are still required to pick, scan, and recover from every workflow exception.
Senior UX Lead — sole UX on workstream
Software, hardware, ops, program
2023 – 2025 · Launched October 2025
After initial launch in 2023, small inefficiencies compounded into measurable operational impact to throughput and inventory quality.
Fast-paced fulfillment centers with handheld devices, time pressure, and constant physical movement.
Frontline associates executing repetitive tasks for 10+ hours a day across rotating shifts.
Hardware limitations, strict performance metrics, and an aggressive launch schedule that challenged a user-centered design process.
Operators were navigating fragmented task flows with limited visibility into system state. The combined friction was preventing the program from achieving General Availability for its 2026 launch dates.
I designed the research to capture both system and human errors at scale, focused on what happens during repetitive workflows that simple usability tests miss.
Rather than focusing on isolated usability issues, I mapped end-to-end task flows across human, device, and backend system interactions — and made the exception branches first-class citizens of the map.
Journey maps from the field were then used in working sessions to educate stakeholders on the current UX — and to gain alignment on feature prioritization for both next-generation hardware and near-term software updates.
Every redesign decision was anchored in a measurable cost. The pick cycle baseline:
The time data pointed to where the cycle was leaking — which task steps consumed disproportionate seconds and were therefore worth redesigning. The qualitative data from contextual interviews explained how each of those steps could actually be improved.
Given fixed hardware and legacy architecture, the redesign followed a strict set of rules — every change had to clarify state, reduce decision complexity under time pressure, and stay within the existing software architecture.
Pick — product identifiers + scanned barcodes
Consolidation — multi-barcode scan
Instead of optimizing only for happy-path tasks, I prioritized designing for failure points — reducing cognitive strain and preventing the downstream rework that compounds across a shift.
Scan error — severity-coded recovery
Problem reporting — explicit recovery path
Design for failure before success. In high-volume operational systems, the cost of an unclear error state is paid every shift, every station, every operator.
I defined quantitative metrics tied to each task-flow step and qualitative interview questions to collect feedback across the four sites — validating the new UI's effectiveness against the same baselines the program was launched on.
Each metric was instrumented before launch — every shipped change had a measurable success signal in production, paired with qualitative interview data from the same four sites.
In October 2025, the new UI for all 11 workflows was released to over 800 stations across four U.S. fulfillment centers.
Reduced task error rate
Improved task completion time
Decreased problem-solve labor hours
Increased user performance rates
The ToteASRS redesign also became the baseline for UI improvements to other workstations with similar workflow tasks — the legacy Pick Station saw a 4% performance lift, and the Inventory Induct Station saw +7% performance with 3.5 hours/day of problem-solve labor eliminated.
Once ToteASRS validated the patterns in production, two other stations adopted the same multi-barcode scan, severity-coded error, and explicit-recovery UI conventions:
Get in touch
Currently exploring senior & staff-level UX roles in systems thinking and service design.