Redesigning How Experts Work with Data, AI, and Systems
Client
International Monetary Fund
Year
2025/2026 Summer
Scope of Work
ENGAGEMENT OVERVIEW
Three scales. Three research approaches.
SYSTEM
Public Finance Systems Modernization
Service Blueprint · Systems Mapping · North Star Visioning
50 fragmented pain points
→ 10 key opportunity areas
Timeline: 2 month (Jul - Aug 2025)
DATA
AI Forecasting Tool Adoption
Usability Testing · Comparative Research · Expert Interviews
Discover AI output Expectation gaps → Adoption Strategy
Timeline: 3 months
(Jun - Aug 2025)
ENTERPRISE
Enterprise Data Discovery
Journey Mapping · Stakeholder Facilitation · Concept Validation
UX assessment to strategic → Implementation action plan
Timeline: 6 weeks (Jun - Jul 2026)
01 — PUBLIC FINANCE SYSTEMS MODERNIZATION
Connecting experience and systems on the service blueprint
Service Design · Cross-agency Systems Mapping ·
The IMF’s public finance modernization team was exploring how modernization could go beyond digitizing legacy systems. We used service design to understand how the existing system worked, where people experienced friction, and what needed to change before defining the future state.
MY CONTRIBUTION
Based on the qualitative data gathered from stakeholder workshops, I mapped the AS-IS budgeting process across 10+ agencies and 7 phases into the service blueprint. It contatins key activities, user pain poitns, data flows, operation system. I translated 50+ user & operaional frictions into 10 actionable modernization opportunities.
Methods: Service Blueprint, Stakeholder Workshop
HOW I APPROACHED IT
Built a traceble structure between experiences and systems
The frictions came from different agencies and stages of the budget cycle, so grouping similar issues alone would not make them actionable. I connected each opportunity to three dimensions of modernization — process, technology, and governance — and mapped it back to the specific frictions it addressed across the service blueprint. This created a traceable structure between what people experienced on the ground and where the broader system needed to change.
OUTCOME
Contributed to the team’s first service-design-led PFM modernization engagement, an approach later piloted across 3 countries and developed into a playbook and broader initiatve.
02 — ENTERPRISE DATA DISCOVERY
Reframing an interface problem as an end-to-end workflow
UX Research · Platform Migration · Enterprise Tool & Workflow
The initial challenge centered on improving search and discoverability within an internal data platform. But research revealed that economists' actual journey extended far beyond the interface.
MY CONTRIBUTION
I led the six-week design sprint end-to-end, from research planning and interviews to journey mapping, 2 stakeholder workshops, concept validation, recommendations, and the final report.
Methods: Design Sprint, Semi-structured Interview, Cognitive Walkthrough, Prototyping Test
THE TURNING POINT
Opening up directions, including firm-wide integrated search and AI-assisted discovery.
The most important friction appeared outside the interface we had been asked to evaluate. I extended the journey before and after the product experience and found fragmented entry points across the broader discovery ecosystem. That changed the problem we brought back to stakeholders. The conversation expanded from improving one interface to improving how economists discover information across the organization, opening up directions including integrated search and AI-assisted discovery.
OUTCOME
Delivered and presented the implementation plan across near- and long-term priorities to cross-functional teams and leadership, providing an actionable direction for the platform’s future improvement. The proposed direction received strong positive feedback from the client team.
03 — AI FORECASTING TOOL ADOPTION
Understanding how AI fits into expert judgment
UX Research ·AI Adoption · Enterprise Data · Pilot Testing
An AI forecasting tool was being piloted with economists working in data-constrained environments. The team needed to understand how the product performed in use and what would influence broader adoption.
MY CONTRIBUTION
I conducted usability walkthroughs and discovery interviews with existing pilot users, new users, and technical experts, comparing expectations across different levels of familiarity and expertise.
Methods : Semi-structured Interview, Contextual Inquiry
WHAT THE COMPARISON REVEALED
Economists weren't simply consuming forecasting data. They were accustomed to building models, interpreting results, and applying professional judgment to produce forecasts.
Introducing AI-generated forecasts created a new gap: where does this evidence fit into a process experts already understand and control? Users needed to understand how the forecasts were generated, what they could adjust, and how AI-generated evidence should inform their own judgment.
Ease of use and willingness to trust the output were two different questions. Comparing perspectives surfaced expectation gaps around what users needed to understand about AI-generated forecasts before incorporating them into their work. The research helped distinguish interface usability from the broader questions of transparency, workflow fit, and trust that shaped adoption.
New technology needs to fit the mental models behind existing expert workflows.
OUTCOME
The findings informed recommendations around transparency, workflow integration, onboarding, and product positioning as the team considered broader adoption.
18 Countries Pilot → 55 Counties Expansion (2025 Winter) → Global Launch (2026 Summer)
Takeaway
When designing for AI adoption, I would start by mapping how experts currently form judgments before introducing a new tool into their workflow.
LESSONS LEARNED
What research looks like when the domain expertise isn’t yours
The scale of the problem should determine the research approach.
A multi-agency system called for service blueprints and systems mapping. An enterprise discovery problem required looking beyond the product through journey mapping and stakeholder workshops. An emerging AI product required comparing usability, expectations, and trust across different types of users.
Research is most valuable when it helps a team decide what to do next. Across these engagements, findings became useful when they helped stakeholders align on the problem, determine what deserved attention, and move toward future-state design, product direction, or an action plan.
Working with domain experts taught me to be clear about the expertise I bring to the room. I couldn’t and didn’t need to match economists or technical experts in their depth of knowledge about economics, data, or modeling. My role was to understand the human and organizational side of the system: how people behave, how work actually happens, where decisions become difficult, and where expectations diverge.
That made me pay closer attention to the gap between what people say they need, what they actually do, and what the organization expects them to do. Staying close to those behaviors and workflows gave me a way to contribute in highly technical environments while keeping the research grounded in the business challenge and the decisions ahead.