Current Vantage Point
I've been working on a human-centered design team in an IT department during the current surge of LLM-powered tools. I've witnessed first-hand the ups and downs of an enterprise attempting to adopt AI tools – the hype, the legal and security due diligence, the excitement, the disillusionment, and the potential finally being realized in pockets.
But we're mostly a "buy over build" shop, so with everything happening in AI and product development right now, the FOMO is real...
I'm currently seeking a highly collaborative team where I can work with engineers and PMs to ship impactful products / services that leverage AI thoughtfully, in ways that maintain human dignity while pushing the boundaries of business productivity and scientific progress.
I'm also looking to continue blending AI tools in my research and design processes. ChatGPT, Claude, and Dovetail's AI analysis features have already accelerated my work. I'm just starting my journey with VS Code Chat and Claude Code for design and development, and I'm thrilled that we're reopening the professional dialogue about designers working with code. We'll always benefit from having engineering-focused counterparts, but everything runs smoother when there are fewer layers of abstraction and translation between us. I've known HTML for 20+ years and have prototyped in code for past client projects. It comes back easily, just like riding a bike. I'm glad I have the foundational skills and mindset (in part from my Information Systems degree) to learn and adapt quickly to any new technology.
AI-Related Projects
A few snapshots of projects involving AI from throughout my career, all the way back to the days when IBM Watson won Jeopardy.
Enterprise AI Prioritization | Travel Chatbot Pilot | IBM Watson for Oncology & Insurance | Health Monitoring Chatbot
Gates Foundation – AI Opportunity Prioritization
When the senior leadership issued a directive to explore AI's potential for transformative process improvement, the foundation's CIO called me in to help, pairing me with a program manager from the COO's office. Leadership had nominated the Travel and HR Talent Acquisition teams as two areas where GenAI advancements had transformative potential.
We quickly got up to speed on where each team was coming from to meet them where they were at. Then I led rapid journey mapping workshops with each team to map their end-to-end process experiences, surfacing pain points and AI opportunity areas, ultimately capturing 38 rough ideas across both teams.
There were more ideas than the teams had capacity for, so we developed idea evaluation rubrics to score concepts for efficiency gains, business value impact, user benefit, and feasibility. Working sessions with each team produced a visualized landscape of opportunities and a leadership-ready narrative on priorities, capacity, and next steps – ultimately landing on 4 priority pilot concepts across the two functions.
We had some very exciting transformative ideas, but "wait and see" was the best move for many of them. In enterprise contexts deeply tied to existing systems like Workday and Concur, sometimes the right AI strategy is patience. Vendors are ramping up their AI features, and for internal efficiency goals without competitive pressure, waiting for something "free" within your existing software can be wiser than rushing into a shiny experiment.
Gates Foundation – "Ask Travel" Chatbot
The Gates Foundation's global travel program is genuinely complex – staff range from first-time conference attendees to program officers spending months in the field. Travel is often to countries with specific business visa requirements and high-risk security considerations. When the Travel team and IT collaborated on an MVP chatbot built in Microsoft Copilot Studio to handle policy, logistics, visa, and travel security questions, my role was to determine whether it was ready to meet that complexity at org-wide scale.
I designed a multi-modal pilot evaluation mixing 1:1 usability sessions, live group drop-in testing, asynchronous testing, and an exit survey, deliberately recruiting real staff rather than relying solely on SME testing. Everyday users phrase questions differently than the travel program experts, and in a probabilistic generative AI system that phrasing variation can make a difference in outcomes.
Across 21 participants, 11 survey responses, and 100 thumbs-up/down feedback comments, I surfaced and helped triage ~22 issues in collaboration with the tech and travel teams. After resolving the top 12 most important and/or easiest issues, the chatbot is launching to all staff spring 2026.
The evaluation framework I developed for perceived accuracy, completeness, usability, and trust is now a reusable resource for the several other internal chatbot projects underway at the Foundation.
Mad*Pow Project – IBM Watson for WellPoint
In the early days of IBM Watson's healthcare ambitions, I designed one of the first interfaces for an AI recommender system built for WellPoint/Anthem, a large health insurance company and claims processor for many Blue Cross Blue Shield affiliates.
Watson ingested patient records and matched them against medical research, clinical guidelines, utilization management policies, and cost data – surfacing ranked recommendations with confidence levels, reasoning, and cited sources.
My design work centered on making those recommendations legible and actionable for two very different users: in-network oncologists exploring treatment options for their patients and utilization management staff reviewing pre-authorization requests for medical necessity. I defined the human-in-the-loop workflows for accepting or overriding Watson's recommendations, surfacing alternate clinical judgments, and providing feedback to train the model over time.
What began as a 2-week prototyping engagement extended to a year-long collaboration based on client satisfaction with my work.
Watson's broader healthcare trajectory was more complicated – results didn't fully match the hype, and errors in clinical recommendations surfaced. Pre-authorization and evidence-based treatment decisions remain a challenge in the industry. I'm optimistic about what modern AI can bring to them, and clear-eyed about what it requires: accuracy, transparency, and safety have to be non-negotiable when people's lives depend on the output. I'd work on this problem again in a heartbeat.
Mad*Pow Project – ImagineCare
ImagineCare, a remote patient monitoring service built from scratch with clinicians at Dartmouth-Hitchcock Medical Center, blended human care team support with ML-powered automation to help patients manage wellness and chronic conditions from home. Sensor data and mobile inputs ran through clinical algorithms enhanced over time with machine learning – flagging anomalies and triggering either automated chat messages or care team action items when human outreach was warranted.
The chat message library for the bot was shaped by design principles defined through activities I facilitated with the clinical product team throughout the design process, as well as a behavior change design workshop facilitated by my colleague and mentor Dustin DiTommaso.
Part of my design work centered on the chat experience – defining interaction patterns and guidelines for bot-to-human handoffs. My user testing revealed that messages with a human face and name felt more motivating, but as a team, we made a principled decision not to disguise automated messages as human. We landed on a shared "ImagineCare Team" avatar for bot-sent messages, preserving authenticity and trust at the risk of potentially losing some engagement.
It's a moment I find myself reflecting on often as AI-powered bots proliferate, and reflects a design philosophy I still hold – because thanks to our holistic design work the service still high achieved engagement, 4x a national benchmark for wellness programs.
Read the full ImagineCare case study or explore all case studies.