As Lead UX Designer at Amazon, I led the design of critical operational tools used by thousands of associates across Amazon's global fulfillment network. I simplified complex workflows, reduced training time, and improved operational efficiency through human-centered design.
Amazon
Lead UX Designer
14 months (redesign of three products + full system integration)
CSAT for STEM product
Associate onboarding time, reduced from 40 hrs
Mis-sort & pallet errors, down from 65%
Sort Center teams juggled four disconnected tools—Resource Management, Resource Allocation, Chute Management, and a legacy handheld—resorting to manual, non-standardized spreadsheet workflows with no validation. Managers spent 20–30 minutes walking the floor to coordinate changes, leading to high error rates, lengthy training, and missed SLAs.
By collapsing these silos into three purpose-built but tightly integrated products, we could:
During the early months of COVID, in-person site visits were impossible, so I got creative to fully understand the user context and true problems:
Virtual tours & live camera feeds
ACES managers ran guided, real-time video walkthroughs of choke-points and high-traffic areas. Observing workflows and equipment layouts remotely ensured my designs reflected actual operational constraints.
Remote user interviews & prototyping
I held weekly video sessions with ACES managers, who recruited shipping clerks, area managers, and associates to test interactive prototypes of STEM, COS, and SAM. This surfaced role-specific pain points and allowed rapid iteration, even without physical floor time.
Archival research
I reviewed pre-COVID field studies from North America and Europe to fill in context gaps—leveraging past observations, equipment configurations, and user feedback to inform hypotheses for new, integrated solutions.
Once lockdown restrictions eased, I resumed on-site visits—spending two days each month in 2021 at the Kent Sort Center shadowing associates and shipping clerks. Seeing a water-spider use the new directed-work feature on SAM in real time drove critical refinements to our task-assignment logic and UI.
By combining these tactics, I bridged the physical divide imposed by the pandemic, gained empathy for both planners and on-floor associates, and ensured my designs tackled genuine, deeply-rooted problems rather than assumptions.
"I have 25 tabs… by the time I'm done, half my rules are stale."
"By the time you make that edit, it's 20 or 30 minutes later and you can't even address it."
"Associates guessed which chute to clear; managers had to verify in person."
I partnered with our engineering lead and ACES managers to launch an Early Adopters Program that spanned NA and EU pilot sites. Over six intense weeks, I facilitated cohort workshops with shipping clerks, area managers, process engineers, and reliability maintenance leads—running contextual inquiries on RM, RA, and CM workflows, and guiding hands-on prototype walkthroughs of early STEM concepts. Together, we mapped every pain point (from modal overload to manual Excel exports) into a consolidated journey map, then convened a North Star Vision workshop to co-define our integrated STEM tool's success criteria—click-path reduction, inline validation, and streamlined hand-offs. By surfacing real user voices, iterating in rapid design sprints, and aligning on tangible KPIs, I broke down adoption barriers, secured cross-functional buy-in, and laid the groundwork for a 95% adoption rate at launch.
At the outset, each tool lived in its own silo—different UIs and mental models. This lead to fragmented experiences for the customers. I crafted different flow diagrams, decision maps and diagrams that distilled these complex workflows for myself while serving as collaborative alignment artifacts with ACES managers, product leadership, and engineering. These shared maps became our single source of truth—anchoring design principles, guiding development priorities, and driving a unified integration roadmap.
RM⇄RA Workflow Convergence
Side-by-side flowcharts of Resource Management (defining equipment) and Resource Allocation (assigning rules) surfaced overlap. We designed a single table-based interface where both tasks co-occur—no tool-hopping or Excel exports.
Embedded Planning Flow
Mapped how planners review forecasts, tweak stacking filters, and preview before/after chute loads. Inline banners now prompt only when a human override is required.
Mapping SAM Functional Tasks
I audited every function in the legacy CMM-UI, diagramming end-to-end workflows for container clearing, tote routing, and error handling. This deep dive surfaced broken logic paths and redundant screens that were confusing associates and slowing task completion.
Designing Role-Based Flows
Partnering with ACES and Operations teams in NA, EU, and JP, I aligned these functions to real-world roles and training states. By mapping each function to internal and external associate roles, I transformed SAM from a generic "choose-your-own-adventure" interface into a streamlined, permission-driven tool—ensuring users saw only the screens and actions they were trained to perform.
Alert-Trigger Flow
Old Model (Taxi):
Associates wandered the floor seeking full containers—like taxis waiting at busy corners.
New Model (Lyft):
SAM's DCM "brain" assigns each associate their next highest-priority container with clear, on-device instructions—minimizing idle time and walking.
Shipping Clerk & Associate Workflow Mapping
I mapped the shipping clerk's end-to-end journey and their interactions with associates, overlaying workflows from STEM, COS, and SAM to pinpoint manual hand-offs—like chute clearance notifications and rule updates—that could be eliminated through integrated, event-driven communication.
By mapping both individual workflows and their interactions, I created a continuous, event-driven ecosystem that eliminates context-switching, manual hand-offs, and stale data.
Three event-driven products—STEM for configuration, embedded COS for planning, and SAM/DCM for execution—work together so that:
The real impact came from designing the flow between tools—not just the tools themselves. Rules set in STEM fed directly into COS shift plans. COS plans became actionable tasks in SAM. And SAM surfaced feedback that traveled back upstream. The result was a connected, continuous system where each tool reinforced the others.
I led targeted, user-driven iterations in each product to address specific pain points:
Table Consolidation:
Combined equipment setup and rule assignment into one interface—no Excel exports.
Bulk Swap & Review:
Multi-rule edits in one action with a concise review modal to catch errors.
Live Forecast Previews:
Inline capacity bars and sparklines under the configuration table.
Dependency Prompts:
Warnings when a change would cascade downstream.
0→1 Handheld Redesign:
Replaced CMM-UI with an accessible, localized, jargon-free app.
Guided Tasks & Exceptions:
Step-by-step workflows with clear reason badges and inline recovery flows.
Through these deliberate, user-centered iterations, I continuously refined each tool's UX, aligning with operational goals and driving measurable improvements in efficiency, accuracy, and satisfaction.
| Metric | Before | After |
|---|---|---|
| Package recirculation | 48% | 26% |
| Configuration time | 2 hrs | 15 mins |
| Associate training time | 40 hrs | 35 hrs |
| Mis-sort & pallet errors | 65% | 10% |
| CSAT for STEM product | — | 95% |
Financial Impact: $3.5M/yr savings on training; $2M/yr savings on exceptions
Recognition at Amazon's All-Hands: SAM's accessible, fully localized redesign was featured at Amazon All-Hands as a model for inclusive, region-adaptable operations tools
"The UI is more intuitive—filters and pop-out details make it a breeze."
"I was skeptical about Chute Optimization, but this is 10× better than expected."
"No more guessing where to place packages—I get clear instructions now."
"This is one of the smoothest transitions to teach someone."
By focusing on clear, table-based configuration, in-context planning, and a guided, role-based execution app—tied together by real-time event flows—we eliminated spreadsheets, manual walks, and guesswork, dramatically boosting efficiency, accuracy, and satisfaction across Amazon's global sortation network.