This project applies usability testing and heuristic review to surface navigation barriers on Washington 211’s resource hub, delivering WCAG-aligned fixes that enable users in crisis to find housing, food, and utility aid in seconds.
Washington 211 - Main Client DSHS Aging & Long-Term Support Administration - Support Site Savvy - Webmaster
Usability Researcher - lead UX Designer
Miro Figma Google Forms
Usability Researchers UX Designers Usability Analyst W211 Board Members
How do radiologists make decisions when the signs aren’t obvious?
When someone close to me was diagnosed with breast cancer, I saw how much uncertainty surrounds the diagnostic process. It made me curious about how design and machine learning could bring more clarity and confidence to real-world diagnosis. That experience stayed with me, and I got the chance to pick the project back up with the University of Washington to explore those possibilities further.
"Misdiagnosis and overdiagnosis remain key challenges in breast cancer imaging, where conventional mammography may fail to detect lesions."
Breakups can feel like grief – sudden, isolating, and overwhelming.
And yet, the digital tools available often emphasize little more than ‘just move on.’ We asked, what would it look like to treat heartbreak not just as pain to ignore, but as an experience to grow from?
"85% of US adults report experiencing a romantic breakup, with 1/3 of those individuals experiencing clinically significant depressive symptoms"
How might we create tools that make cancer diagnostic data more interpretable, transparent, and actionable for radiologists?
How might we create a digital experience that adapts to the psychological realities of breakup recovery, including attachment styles, identity loss, and emotional dysregulation, while remaining clinically grounded and deeply human?
We started by looking into existing research to understand why diagnosing breast cancer is often so complex. Radiologists interpret features like shape, margin, and density differently, and even small changes can lead to different outcomes. This is especially true in borderline cases or when images aren’t clear. These insights helped us focus on where and why the problem exists and to design tools that support clinical judgment.
The machine learning model was built early on, during my time at University of Nottingham, as a way to explore how tumor characteristics could predict malignancy and patterns, especially whether those patterns aligned with how radiologists make decisions. At that stage, I didn’t know this would evolve into a design project. But training the model helped uncover which features were most influential, which later became critical input for designing an interface that could surface meaningful, case-specific insights and support clinical reasoning.
After deciding to turn this into a design project, I revisited the model through deeper quantitative analysis to unpack how its predictions worked in detail. I wanted to explore which features to emphasize, how uncertainty showed up in the data, and where edge cases might cause confusion. These visualizations helped shape the design direction, especially around what to prioritize, how to handle ambiguity, and how to build trust through clarity and transparency.
I started with secondary and market research to see if this problem was just mine or something more people were going through. I wanted to understand how common breakups are, how deeply they affect people, and whether there were any patterns in how we experience or cope with them. I also wanted to identify where the gaps were – both in how breakups are talked about in the community and existing tools that are actually out there.
I conducted a competitive analysis to get a clearer picture of what breakup and emotional recovery tools are already out there. I wanted to see how others were approaching the problem, where they were falling short, and how Repose could stand out. It also helped me spot patterns, such as how most apps lean on generic, self-guided content that doesn’t really adapt to users’ emotional needs. I looked at things like usability, content tone, and how these apps try to keep people engaged. This gave me a better sense of what’s working, what feels impersonal, and where people might be looking for something deeper. It set the foundation for designing something more intentional and emotionally supportive.
After the competitive analysis and literature review, I wanted to understand what people were actually experiencing after a breakup, but at a larger scale. I designed a survey to capture broad, real-world insights into the emotional challenges people face. Surveys gave me the reach to validate early breakup patterns, hear from a diverse group of people, and identify key segments that could inform future design decisions.
We included an exclusion criterion since emotional experiences tend to fade or shift over time. This helped ensure the insights were grounded in more recent, emotionally relevant experiences.
The findings validated a real, unmet need for breakup support that feels personal, responsive, and human – shaping the foundation of our product vision.
After hearing from dozens of people through our initial survey, we invited those who expressed interest in a follow-up to take part in interviews. Using convenience sampling, we spoke with 12 participants who had recently gone through a breakup within the past two years. These one-on-one interviews helped us go deeper into their emotional journeys, uncovering nuanced needs and pain points that broader surveys couldn’t capture. Speaking directly with them gave us a more intimate understanding of why existing tools often fall short and the type of meaningful support could actually look like.
8 key findings. Raw, emotional, high impact.
They revealed how breakups disrupt everyday life, emotional stability, and the need for compassion, community and support. These insights shaped Repose as a focused breakup recovery tool, designed to offer structure, emotional guidance, and self-directed healing when support feels out of reach.
Many participants described the support they found, whether from apps, friends, or articles, as too generic or emotionally shallow. They wanted layered guidance that matched the depth of what they were feeling.
People shared that recovery did not follow a straight path. Triggers showed up unexpectedly, and emotional needs shifted over time. This revealed the importance of having support that can adjust with them.
What people valued most was not just being told what to do, but feeling understood. Tools that reflected their experience or offered emotional validation were seen as more helpful than ones that were purely solution-oriented.
Before designing the interface, I needed to understand how diagnostic decisions break down – in models, in data, and in clinical workflows. I triangulated three methods to create a research-to-design matrix that helped validate patterns across sources and identify high-confidence insights. The matrix surfaced ten core findings that revealed where errors happen, what users actually need, and how AI predictions can be made more interpretable. These insights became the foundation for every UI decision that followed.
3 methods; 10 findings; 4 that deeply shaped the interface.
By combining model behavior, pattern analysis, and literature on breast cancer diagnostic processes, I mapped out the most impactful pain points: unreliable feature weighting, lack of transparency, edge cases, and cognitive overload. Each design decision below directly addresses these breakdowns with targeted interface responses.
Radiologists don’t just look at a tumor’s ‘average’ size, they also zero in on its single most abnormal spot in a single patient. Missing that one extreme region can lead to under-diagnosis.
"Existing breast imaging studies reported the entropy, mean, minimum, and maximum as important features."
Breakup recovery isn’t linear, so we needed a way to understand the ups and downs people go through. We pulled insights from research, surveys, interviews, and other tools to map out how people’s emotions and needs shift over time. Seeing it laid out like this helped us understand when people feel most overwhelmed or unsupported, and it made it clearer what Repose should offer and at what stage.
We heard over and over that breakups left people feeling stuck. More than comfort, they wanted clarity, momentum, and reassurance that they’d be okay. To organize these needs in a way that could guide design decisions, we used the Jobs to Be Done (JTBD) framework. It helped us see each pain point as part of a larger goal and gave us a clearer sense of how Repose could actually support real progress.
This approach also gave us a strong foundation for identifying our value proposition and making decisions about what to build first. It pushed us to think beyond about what people feel, but what they would actually use, need, and pay for.
“When I feel like I wasn’t enough, I want to hear from others who’ve been through it, so I don’t feel broken and alone.”
“When I keep replaying the breakup, I want to find clarity to make sense of what happened, so I can stop fixating and spiraling.”
“When I’m overwhelmed and panicked, I want to feel emotional relief by grounding myself, so I can get through the day.”
These jobs became design anchors that clarified what we needed to support and why. They helped us prioritize features like guided self-reflection and content for regaining identity. Instead of jumping straight to features, we grounded our design direction in helping users make real emotional progress.
I co-built the interface using Streamlit and iterated directly in code (using vibe-coding), guided by user needs and model behavior. Streamlit allowed me to maintain full control over the model logic while rapidly prototyping interfaces that stayed true to the algorithm’s outputs. Unlike visual design tools, Streamlit let me directly connect model predictions with interface elements, making it easier to test ideas in real time, adjust how probabilities were framed, surface uncertainty, and experiment with interactive features like sliders, graphs, and confidence estimates.
To ensure Repose was viable and grounded in real user needs, we created a Business Model Canvas and mapped out our customer segments early on. This helped us define our core value, identify key audiences, and clarify how we would reach and support them. These insights gave us the confidence to move forward with a focused MVP, ensuring our solution remained aligned with both user needs and business goals.
To ensure Repose was viable and grounded in real user needs, we created a Business Model Canvas and mapped out our customer segments early on. This helped us define our core value, identify key audiences, and clarify how we would reach and support them. These insights gave us the confidence to move forward with a focused MVP, ensuring our solution remained aligned with both user needs and business goals.
A step-by-step welcome flow that gathers your companion, attachment style, and personality to tailor every lesson and reminder to your needs.
Greets users with a warm introduction, highlights core benefits, and invites them to begin their healing journey
Prompts users to choose a plant or animal companion, immediately tailoring the journey and building an emotional bond
Asks about your typical relationship pattern to customize lessons and exercises around attachment needs
Captures personality/social preference (introvert, extrovert, ambivert) to adjust how and when reminders and content are delivered
A bottom tab bar granting instant access to your daily tools, guided micro-lessons, community discussions, and profile settings for a seamless healing experience.
Central dashboard for daily coping tools all in one glance – breathing, affirmations, journaling, and habit tracking
Bite-sized, personalized audio lessons organized by topic and healing stage, with clear duration and play controls
Anonymous forum where users can browse or join discussions on common breakup challenges, share stories, and find peer support
User settings hub: swap your healing buddy, set daily check-in reminders, and review personal data to keep the experience tailored to you
A central dashboard of mindfulness exercises – breathing, affirmations, habit tracking, and journaling – designed to help you stay present and build healthy routines.
A guided, timed breathing exercise with a simple countdown to help users calm their nervous system and reduce anxiety.
Series of positive, self-compassion statements presented on screen or via voice to counter negative thoughts and boost mindset.
Daily checklist for small, customizable actions (e.g. drink water, take a walk) that encourages building consistency through streaks.
Dual-mode journaling (free-form mood journal or prompt-based session) for users to reflect on feelings, track patterns, and gain insight.
Qualitative Research Methods | Customer Segmentation & JTBD Mapping | Research-Strategy Translation
Quantitative Methods | Machine Learning & Data Visualization | AI Based Prototyping
