Amazon’s internal ML models were managed through a tool called DasBoard, owned by the Modeling and Optimization Platform (MOP) org. After four years of engineering-led development, the product was fragmented, inconsistent, and hard to scale. I was brought in to lead the UX redesign to simplify workflows, improve model usability, and set a foundation for future self-service growth.
Information architecture created to restructure DasBoard for clarity, scale, and reusable workflows.
Led UX as the solo designer on DasBoard, driving platform direction and execution
Introduced North Star framing and design principles to guide long-term product thinking
Coached engineers on UX patterns and building scalable, reusable components
Shared dashboard patterns with the Meridian Design System team to uplift org-wide quality
Translated complex ML workflows into usable, self-service interfaces across teams
Four years of engineering-led development created fragmented IA and deep UX debt
Inconsistent patterns across teams made similar workflows feel disconnected and hard to scale
Most engineers and leaders had never worked with a designer, requiring cultural alignment and upfront trust-building
No product manager support and limited engineering capacity meant prioritization and direction had to be self-led
Original DasBoard interface, created without UX support and focused on dense, engineering-driven layouts.
A key screen redesigned to improve clarity, hierarchy, and usability for ML run data and outputs.
Delivered a modular UX framework that simplified job creation, reduced redundancy, and improved flow clarity
Shifted engineering focus from one-off features to reusable components and scalable platform thinking
Raised design quality across tools by contributing core patterns back to the Meridian Design System team
Improved usability for data-heavy tools by refining how ML results, exceptions, and task logic were presented
Enabled the team to scale with less design input by anchoring in repeatable patterns and interaction models
Interviewed DS teams to surface workflow pain points and clarify key exception paths
Validated design direction with both stakeholders and end users through iterative reviews
Audited internal tools and gathered technical feedback to align patterns with backend architecture
Reworked IA to eliminate duplication and organize tasks by lifecycle stages
Simplified job setup and result review while reducing surface area across exception flows
Prioritized foundational flows and deferred advanced features like mapping, ensuring future additions wouldn't require rework
Delivered modular, high-fidelity flows structured for phased rollout and reusability
Anchored visual and interaction patterns in principles of transparency and self-service
Worked closely with engineers to vet technical constraints and ensure UX decisions were feasible
Led working sessions to align engineering on platform priorities and phased MVP delivery
Shared reusable dashboard components with the Meridian Design System team to support broader org needs
Helped shift the team's mindset from feature delivery to platform consistency and long-term scale
Redesigned job type flow guiding users through complex setup with a simplified, step-by-step structure.
Beyond DasBoard, I supported UX maturity across the MOP org through design ops, education, and team advising:
Hosted UX office hours and served as a design resource for adjacent engineering and data science teams
Reshaped how ML model outputs were presented, making insights more usable for non-experts
Shared reusable patterns and component logic with the Meridian Design System team
Advised teams across Amazon Ops who were building dashboards without embedded UX, helping align tools under shared principles
