.png)
11.6% of the U.S. population diagnosed with diabetes. Our research discovered that there is a missing link between patients' blood glucose levels and contextual data. This resulted in GlucoSense, a holistic health-tracking app for diabetes management, powered by AI.
Time: Jan to June 2024
‍Role: Research, Strategy & Design
‍Sponsor: Microsoft
Type: AI / App Design
.png)
Our project was selected as the Capstone Award for Excellence in Innovation, here's what I've accomplished:
‍
• Conducted 6 expert and 3 patient interviews using my background as a UX researcher, uncovering insights that shaped the solution. Â
• Designed high-fidelity, clickable prototypes and mapped onboarding flows to showcase the platform's key functionalities. Â
• Led a co-design workshop to create features mitigating AI risks, prioritizing patient safety and well-being.
Design Challenge
How can we improve patient experience through the use of AI?
Project Goal
Establish a guiding vision for AI-driven health tools
‍‍
We found that during the short 30 minutes of consultations for most out-patients, doctors often ask patients to recall the context behind when certain "crash-events" (aka extreme low or high blood glucose level) happened so that they can better understand what led to these events and recommend better-informed health suggestions to patients. However, as humans, we can barely recall what we ate this morning.
Since September, I've been a Product Designer at Tether Data, a startup creating a planning and inventory automation tool for e-commerce brands. Collaborating with the PM and founder, I transformed low-fidelity wireframes into a high-fidelity clickable MVP to attract potential customers and developed a lightweight design system.
Time: Sep 2023 to Current
‍Role: Product Design
‍Client: Tether Data
Type: Enterprise Software
We reached out to 30 endocrinologists at UW Medicine and spoke with six of them, recruiting patients to validate the problem space. Additionally, we sought directional advice on approaching AI tools by meeting with the Chief Medical Officer and Vice President of Healthcare at Microsoft.
As the first designer on the team, here's what I've accomplished:
‍
‍• Transformed low-fidelity wireframes for the entire product into a high-fidelity, clickable MVP with over 200 screens, successfully attracting interest from four potential customers.
• Worked cross functionally to connect with a PM and stakeholders to translate research and user stories into mockups
• Built a lightweight design system with reusable components on top of shadcn/ui
IxDA is a volunteer-run organization that facilitates social and educational events for designers in Seattle. The objective of this project is to design a landing page for IxDA. The design goal is to establish a stronger connection to the organization's physical location and clearly communicate its mission.
In this project, I designed the logo, landing page, and marketing materials.
Time: March - June 2024
‍Role: Visual Design
Type: Web Layout Design / Branding
The main design goal was to enhance the organization's identity by emphasizing the human element within it.
It was important for the website to effectively convey that the organization regularly holds social and educational events in person.
Highlighting the physical location of where the organization is based on.
Using my mood board as a reference guide, I developed three distinct logo concepts, each following a strategic approach based on the logo design styles: wordmark, pictorial, and emblem.
I created a moodboard based on the keyword "human experience," inspired by the idea of breathing and air—fundamental elements of all living things. This inspiration shaped the visual theme, emphasizing qualities of being natural, cool, and intangible.
We reached out to 30 endocrinologists at UW Medicine and spoke with six of them, recruiting patients to validate the problem space. Additionally, we sought directional advice on approaching AI tools by meeting with the Chief Medical Officer and Vice President of Healthcare at Microsoft.
"Hesitant to let patients base insulin dosing decisions on AI suggestions due to the many variables involved. Without considering the specific context of their blood sugar levels, it becomes challenging to make treatment decisions without physician supervision."
"I often arrive [to the appointmnet] unprepared, needing to gather information about what I ate, my blood sugar levels, and how much insulin I need. Sometimes, I don’t manage to get the data ready at all. I know I need to do better, but it’s really hard. With everything else happening in my life, manually tracking and taking notes on the past three months feels impossible."
"Consider multimodal data AI and passive ways to capture [patient] data ...the less you can ask the patient, the greater the adoption."
I designed every page for the first MVP. Here's what the product includes:
‍
Dashboard - Provides live updates on all sales and inventory management data.
‍
Demand Plan - The core feature, offering real-time forecasts compared to actual demand across all channels.
‍
Accounts, Promotions, Orders, Suppliers, and Warehouse - Deliver similar data, each tailored to their specific perspective.
I began with the Dashboard, the default screen, aiming to present all essential information without overwhelming the user while offering flexibility in how they view the data. For example, metric cards and data visualizations make information quickly scannable, while filters for specific values enable users to dive into detailed insights when needed.
%201.svg)
It became clear to us that risk mitigation is most important since we are applying emerging technology with uncertain potential harms in a domain where there is no room for error. To address this, we conducted a Responsible AI Impact Assessment, adapted from Microsoft’s Impact Assessment and grounded in Microsoft’s AI principles.
Our main goal in the co-creation workshop was to uncover any remaining concerns, which led to a major insight: users were optimistic towards AI tools yet concerned that it will increase notification fatigue.
.svg)
.png)
I rediscovered the power of storytelling through this project, which played a major role in helping us move our work forward—from convincing clients to support our mission to presenting to our classmates, who had no idea what diabetic users experience day to day. A solid narrative infused with empathy enabled us to engage our audience through numerous presentations and discussions over the course of six months.
Although this product was envisioned as a North Star project shaping the future of AI healthcare and not immediately planned for release, our team approached regulations with seriousness. We asked questions, such as, "How might we integrate this communication loop via the EHR system to enable doctors to easily access data and provide modifications if necessary?"
Having zero knowledge about AI before taking on this challenge, I learned a great deal about the capabilities of AI tools and the real, serious consequences they could lead to when designed carelessly. It made me realize the crucial role designers play and deepened my passion for creating AI-powered tools.
Dynamic Dashboard
In the low-fi wireframes I was tasked with enhancing, every UI element was static. I proposed a dynamic dashboard that encourages users to interact, enabling them to consume information quickly and effectively. This design utilized metric cards and combined similar data visualization cards into a collapsible slider for better accessibility.
Presenting Multiple Layers of Data
Through our daily design discussions, I identified that users prioritize quickly scanning essential data-driven insights while needing the ability to filter and view data by specific warehouses, channels, or products for swift decision-making. To streamline this experience, I implemented a tree dropdown menu for Channels, Warehouses, and Products.
Our Unique Selling Point
Unlike spreadsheets, our "Grouping" feature enables users to filter data based on their preferences effortlessly, eliminating the need to create complex formulas or tables.
After users select the values they want to group data by—such as grouping by Product—a hierarchical data table is generated. This table allows users to expand down to the individual SKU level for detailed insights.
.svg)
• Working closely with engineering to ensure designs are feasible and align with technical requirements.
‍
• Next on my list is revamping the navigation. As we added new pages during the design process, the current navigation system has become less efficient and user-friendly.
‍
• A user onboarding flow is essential to guide users, especially since many will be transitioning from manually entering data in spreadsheets to using this platform.
I spent 10 weeks refining the problem space, researching, testing, and designing a prototype for BWB's mobile experience, addressing the client's original concern on complex parameter entry task and resolving a newly discovered need on status monitoring.
Time: Jan to Feb 2024
‍Role: Research & Design
‍Client: Biodepot LLC
Type: App Design
.svg)
I joined the team as an intern to develop the mobile version of the desktop software.
‍
• Quickly gained domain knowledge and familiarized myself with the software's technical framework.
• To translate the existing experience while introducing a novel way to monitor progress and access workflow details, including logs.
• Advocated for UX and research-driven design, emphasizing the value of iterative design backed by evidence.
Problem
The team was unsure how to translate this complicated task of entering various parameter entries into a seamless mobile experience.
Solution
App with Parameter and Progress Features‍‍
.svg)
I interviewed 3 genomic researchers, each with over 20 years of experience at the Fred Hutch Cancer Center to validate the problem space and discover any unforeseen issues.
-p-500-1.png)
I organized the current flow and mapped out the model flow to make sure our solution tackles the pain points we identified earlier.
-1.png)
The design system became a priority, as the team had only focused on the technical aspects of product development.
.png)
.png)
.png)