As the first and only designer embedded in Modeling and Optimization Platform (MOP) is a data science org responsible for machine learning models that support Amazon’s logistics systems, from cart placement to final delivery. Within MOP, DasBoard was a self-service tool that allowed data scientists to launch their models and enable internal users to run them with custom inputs like workforce data or CSV files.
The product was highly technical and had been built without design support. This led to inconsistent workflows, conflicting patterns, and interfaces that were difficult for operational users to navigate.
Site map of the proposed DasBoard navigation
First design hire embedded on the DasBoard team
Introduced North Star thinking and design principles to guide product direction
Coached engineers on UX consistency and how to build reusable patterns
Partnered with leadership to align on scalable interaction models across features
Shared new dashboard patterns back to the central TransOps design system
Helped translate academic and technical outputs into more usable interfaces
Four years of engineering-led development created deep UX debt
Inconsistent patterns made similar workflows feel disconnected
Most engineers and leaders had never worked with a designer
What the designs looked like when I started on the project
Mockups of the one of the key screens, Run Details
Introduced a foundational UX strategy that allowed the team to scale without relying on constant design input
Created reusable patterns to reduce redundancy and align interactions
Shifted internal culture toward more user-centered thinking
Contributed back to the broader TransOps design system to raise design quality across teams
Improved usability of data-heavy tools by reworking how ML outputs were presented
Mockups of another key screen with a extremely technical workflow workflow
Beyond DasBoard, I influenced design maturity across MOP by hosting cross-team office hours and serving as a design resource for adjacent engineering and science teams. I helped reshape how model outputs were presented, making technical insights more approachable for non-experts. I also shared patterns and recommendations back to the larger design system, helping align design work across Amazon Operations.
