This initiative centered on building a secure, branded suite of internal Generative AI tools designed to empower teams across content, support, and product strategy. My goal was to create intuitive interfaces for high-performing AI capabilities—while ensuring outputs reflected our brand voice, tone, and domain knowledge.
I led the UX and frontend implementation, beginning with user discovery sessions to identify high-impact use cases. We prioritized tools such as a guided content writer, structured prompt composer, image generator, and brand voice transformer.
To ensure AI-generated content was aligned with internal style and expertise, I implemented grounded retrieval-based systems using Microsoft Azure AI + SharePoint. I curated a structured knowledge base of internal documentation (tone guides, product FAQs, policy docs), which the LLM could reference during generation, improving factual accuracy and brand consistency.
I also developed custom prompt frameworks to shape tone, structure, and audience intent — turning vague user ideas into high-quality outputs. Users could choose between different voices (e.g., “formal,” “casual,” “technical”) which influenced tone tokens and post-processing behavior.
The frontend was built to feel light, focused, and intuitive. Using Figma for design and lightweight frameworks for implementation, I crafted an interface that encourages experimentation without overwhelming users.
The tools are now used daily across the organization, generating everything from internal reports to customer emails to pitch decks — saving teams hours per week and raising overall output quality.
In this project, I was tasked with designing a dashboard to visualize complex healthcare quality metrics, helping medical professionals quickly identify patterns in discharge data.
I began with user journey mapping and stakeholder interviews to understand the different personas interacting with the data — from care coordinators to C-suite executives. Through workshops in Miro, we uncovered pain points related to data overload and visual misinterpretation.
I applied principles of information hierarchy to design modular D3-based visualizations, balancing depth with clarity. Iterative prototyping in Figma allowed for quick feedback cycles with real users, ensuring accessibility and data legibility were prioritized. Tableau prototypes were also used for low-code validation with business users before final D3 implementation.
This product aimed to streamline patient handoffs during hospital discharge — a high-risk phase for readmissions.
We kicked off with contextual inquiries with discharge nurses and case managers to understand pain points. Insights showed that most errors stemmed from poor documentation handoffs and inaccessible action plans.
I created wireframes in Sketch and co-created transition journey maps in Miro. Based on mobile vs desktop usage contexts, I crafted responsive UIs in Angular and Ionic, tuned for situational awareness and quick data entry.
The design system evolved into a full enterprise style guide, unifying the brand and improving maintainability across web and mobile platforms.
This internal tool was designed to let researchers rapidly build and deploy visual stimuli for A/B testing and preference studies (powered by out AI-powered evolutionary algorithm).
I began with rapid whiteboard sketches alongside researchers to understand workflow inefficiencies. Their biggest frustrations centered around clunky input forms and lack of WYSIWYG layout previews.
The final design favored drag-and-drop controls and live previews, drastically reducing setup time. I implemented these using Backbone.js and custom jQuery components, ensuring the app remained lightweight and performant for repeated studies.
By focusing on user empathy and co-creation, we transformed an error-prone legacy system into a modular interface with immediate impact on research speed.
This consumer-facing app sought to modernize matchmaking by emphasizing personality over profiles.
I initiated the design process by mapping competitor flows and conducting user interviews with real dating app users to identify patterns of fatigue, mistrust, and vanity-driven interactions.
I designed a unique “question-first” matching system — users interact with a dynamic quiz UI rather than profiles — promoting authentic discovery over visual swiping.
Built in Angular/Ionic, the mobile UI uses soft gradients, calming motion design, and subtle haptics to evoke a cozy yet trustworthy feeling. This project gave me full-stack ownership of the mobile experience, from research synthesis to pixel-perfect build.