Launching and Scaling Project Umbrella at Amazon

Certain details in this portfolio have been omitted due to Amazon confidentiality. The work shown is intentionally presented as mock files rather than the actual launched products.

I led the UX strategy and design for Project Umbrella, which launched in 2018 as a standardized training and coaching platform integrated directly into warehouse workstations. The platform delivers in-the-moment training, real-time coaching, and performance feedback in the flow of work to help new employees learn faster, improve work quality, and significantly reduce the time required from in-person trainers. The system enables consistent, scalable learning and coaching across warehouse operations while driving measurable improvements in productivity, quality, and operational cost.

My Role

I served as the UX Design Lead for Project Umbrella from early concept through research, pilot validation, and platform evolution. I owned the end to end experience strategy across Associates, Trainers, and Managers. I partnered closely with UX Research, Learning, Engineering, and Operations leaders to define product direction, validate training and coaching modalities, and ensure real world feasibility inside production Fulfillment Centers.

Customer Insights & Ideation

I worked backwards from the customer by conducting deep field research across multiple Fulfillment Centers. I observed Associates performing live warehouse picking and inventory handling tasks, interviewed Learning Managers and Trainers, and personally completed Day 1 operations training to fully understand the onboarding experience. I conducted two full days of direct workstation observation with six Associates processing real customer orders. I also partnered with Learning teams to understand existing training gaps, operational constraints, and content delivery challenges. This research led me to define four core user types consisting of New Associates, Veteran Associates, Support roles such as Trainers and Coaches, and Management. This structure became foundational to aligning stakeholders around the right product scope.

Using these research insights, I created five storyboards and mapped complex end-to-end user flows to align leadership on how learning and coaching could seamlessly integrate into real production work. I explored multiple instructional formats including audio, video, synthesized speech, and headset delivery. I also tested hands-free voice interaction for real-time coaching to allow Associates to request help without disrupting their physical workflow.

Experience Strategy & Vision

I proposed and led the vision for task-based learning embedded directly into warehouse workstations. My strategy focused on training Associates in real context while they performed real work rather than separating learning from production. I designed an experience that combined short-form learning, real-time error detection, and immediate coaching feedback into a single system. I defined how Associates would receive on-demand guidance, how Trainers would coach through a centralized dashboard, and how Managers would receive live coaching alerts on handheld devices. I also shaped the self-directed training experience used by Trainers and Managers to deliver standardized learning content across sites and regions. This vision became the foundation of a unified platform that now supports face-to-face coaching, workstation-delivered coaching, and automated workstation-based training at scale.

Planning & Scope Definition

I partnered closely with Product, Learning, Engineering, and Operations teams to define the pilot scope, rollout plan, and success metrics. I helped plan the first live pilot at a production warehouse using new hires on Day 1. I worked through detailed operational constraints including workstation setup, training inventory, and workflow timing. I also helped establish the learning evaluation framework, defining how we would measure training satisfaction, knowledge retention, productivity, quality, and the reduction of in-person training time. This ensured the experience design remained tightly connected to business and operational outcomes.

Design Execution & Validation

I designed the Associate training interfaces, which had to visually and functionally overlay the existing station software UI, using an established design system. Due to strict technical constraints, no changes could be made to the core station software, so I focused on creating a seamless visual hierarchy and interaction flow between the two overlapping systems. In parallel, I designed the Trainer and Manager dashboards for Kindle, optimized for mobility, quick scanning, and operational decision-making.

I led four rounds of formative usability testing to validate content format, delivery methods, and interaction patterns. This included testing synthesized voice vs. human narration, headset vs. open-speaker audio, and push vs. pull learning models. I returned onsite to test concepts directly with Associates in live production environments, gathering real-time feedback. Findings showed that context-based, in-the-moment training using real work scenarios drove faster learning and stronger retention than classroom-style instruction. Short, glanceable content consistently outperformed long, text-heavy guidance.

I used this data to influence Learning leadership to shift from front-loaded training to embedded, just-in-time guidance. I also invited cross-functional partners to observe usability testing in person, which significantly accelerated alignment and built trust across teams.

Leadership

I regularly presented work to Product, Engineering, Learning, and Operations leadership to build alignment and drive execution. I used customer evidence, field data, and experience storytelling to guide decision making. I coached partners across disciplines to adopt iterative testing and user-centered validation as part of the product development process. My leadership helped establish UX as a strategic driver rather than a downstream execution function.

Impact

The platform launched as a unified training and coaching system used directly within live warehouse operations. It transformed a traditional classroom model where one Trainer supported up to five Associates over two weeks into an in-context learning model where new Associates became productive in just two days. This shift freed Trainers to focus on higher-priority needs and on-demand support rather than prolonged classroom instruction. The system automated training records, access permissions, and content delivery across multiple locations while enabling same-day coaching instead of delayed next-day feedback. Just as importantly, it created a more supportive, scalable, and consistent learning experience for Associates in their first days on the job.

Reflection

If I were to revisit this project, I would invest even earlier in accessibility testing for voice and audio interactions in high-noise environments to accelerate adoption with greater confidence. I would also be more intentional about explicitly designing for a support-not-blame UX philosophy from the very start. Beyond improving performance, the platform helped shift the mental model of training from correction and fault-finding to care, prevention, and inclusion. By framing guidance around safety, ergonomics, and real-time support rather than penalty, we encouraged healthier behaviors, reduced injury risk, and built greater trust between Associates and Trainers. I would further prioritize earlier planning for global scalability, particularly around language support, device availability, and regional operational differences.