LastMinute
Category:
B2B SaaS; Dashboard | Healthcare | Workforce Management
At the time I came onto this project, LastMinute had an MVP that could not support real hospital operations. One manager-facing scheduling function was all they had, and it broke under the actual complexity of multi-unit clinical staffing. Working with a CTO, three engineers, and a design lead, I ran stakeholder interviews and rapid iterative user testing to surface what hospital managers and part-time nurses each needed, and where those needs conflicted. By the end of the project, a single-function demo that couldn't close a hospital partnership had become a two-sided workforce management platform, and the version that secured LastMinute's collaboration with Western Medicine Hospital.
Feature 1: Call Out Tracking
AI-powered absence detection and coverage automation — the moment a shift falls through, the system triggers a replacement workflow and surfaces the best-fit nurses instantly.
Feature 2: Customized Shift
Build customized shifts for complex coverage needs — split one shift across multiple locations, or assign different nurses to cover different time blocks within the same shift.
Feature 3: Availability Preference
Keep staff availability current and accurate — giving managers structured input data to distribute shifts more efficiently.
Feature 4: Shift Swaps
Let staff trade shifts across teams while maintaining credential and coverage rules — without requiring manager sign-off on every exchange.
My Role:
Product Designer
Duration
05.2025 - 11.2025
Tool:
Figma; Linear; Claude
Team:
CEO
CTO
Good is not the starting point.
When I first opened the existing product, I knew immediately that the problem wasn't what was missing. It was what was broken.
The team was ready to move forward and start designing new features. But I stopped. The interface had structural failures that no amount of new design would fix if we didn't address them first. I mapped out every place where the product was asking a hospital manager to make a high-stakes decision on incomplete or misleading information, and brought it back to the team before a single wireframe was drawn. We redesigned from the diagnosis, not from the surface.
That audit changed how I approach every redesign. Good enough visuals on top of a broken structure is still a broken product. Before I reach for solutions, I look for what is actually failing and why. The standard I hold is not whether it looks better. It is whether a user can make a confident decision with it.
100%
Client Satisfaction
0 → 1
From Broken Demo
A non-functional MVP rebuilt into a two-sided workforce platform — validated by real clinical users, recognized by the 2026 iF Design Award, and deployed in partnership with Northwestern Medicine Hospital.
Designing for Clinical Confidence in Healthcare Workforce
I find the gap. I fill it.
Stakeholder Interviews · User Journey Map · Opportunity Mapping
When this project started, the team's focus was clear: design the manager-facing scheduling dashboard. The nurse side was acknowledged but unassigned. As I mapped the full user journey, I realized the product had a missing half. Part-time nurses needed to see, claim, and manage shifts from their phones. Nobody had a desktop at the bedside. I proposed a mobile nurse product line, went back to our nurse end user to confirm the direction, and she validated it without hesitation. From that point, I owned the nurse product end to end.
Move fast. Test faster.
Lo-fi Sketch · Wireframe · Testing · Iterations
As discovery progressed, we had one month to deliver a product that had never been built before. I worked in tight loops: sketch a flow, put it in front of a real nurse user, revise based on what I heard, then move to the next flow. When the first round of testing showed that the card system I borrowed from the manager side was too complex for nurses on the go, I stripped it back to clean branding colors and rebuilt the interaction from scratch. Two validated flows later, I had a clearer picture of what the product still needed than any internal review could have given me.
I don't dismiss edge cases. I investigate them.
Secondary Research · Opportunity Mapping · Prototype
Once the core flows were in place, a nurse user raised something I hadn't planned for: she wanted to tell the manager when she was available, not just receive whatever shifts came her way. My first instinct was to flag it as an edge case and move on. Instead, I ran secondary research and found this wasn't an edge case at all. Self-scheduling is an emerging model in hospital workforce management, where nurses actively set their availability and managers use that data to distribute shifts. I brought the findings back to the team, and we built the availability request feature grounded in real industry practice.
Handoff is a design decision.
Component Library · Annotation · Developer Handoff
Once the design was done, I built the full handoff documentation — version numbers, component states, annotations, and flow markups, all in one place. Engineers went from checking screens one by one to reviewing the full flow once and starting to build. No back-and-forth. This was the first time I'd delivered a handoff at this standard. What I learned: documentation isn't a formality after the design. It's the last design decision you make.
"She reframed clinical scheduling as a dual-stakeholder problem, designing for hospital efficiency and clinician autonomy at the same time without sacrificing either."
Jia
Founder, VSDesign
Other Projects
