Everything I have worked on that I can talk about publicly, grouped by the kind of problem it was.
Judging whether AI systems are actually good, and being clear about what the evidence cannot carry.
Evaluation of two RAG-backed AI chat surfaces at TOPdesk: a self-service assistant and an operator-facing Copilot. Four recorded sessions, 147 answers judged case by case with a stated reason on every verdict, clustered into a root-cause taxonomy and fifteen findings ordered by how much evidence sits behind each.
View Case StudyRigorous evaluation methodology for AI-powered nutrition advice, combining open coding of 100+ traces, custom error taxonomies, and systematic evaluation comparison.
View Case StudyStudies and frameworks that changed what got built.
Comprehensive user research mapping opportunities in incident management workflows through interviews, affinity mapping, and journey mapping with 10 participants.
View Case StudyMixed-methods study across four methods — survey (n=31), card sorting (n=5), concept testing (n=7) and usability testing (n=19) — on how knowledge managers structure a knowledge base and who they need to restrict access for.
View Case StudyReadiness and maturity framework benchmarked against Gartner, Ivanti, HDI, SDI, and Forrester — mapping where customers stand on the path to intelligent service delivery, informing product strategy, AI roadmap priorities, and long-term platform vision at TOPdesk.
View Case StudyGlobally standardised customer maturity scoring program mapping product adoption across ~4,900 customers and 11 regional branches — informing product strategy and the AI roadmap for intelligent service delivery.
View Case StudyA Wharton Strategic Management course project applying the Connected Strategy framework to mid-market ITSM: customer journey mapping, pain-point and information-flow analysis, and four strategy archetypes mapped onto AI-driven automation. Figures throughout are modelled targets, not measured outcomes.
View Case StudyThe tools and playbooks that let research reach decisions without a researcher in the middle of every one.
A copilot that lets any product role run real research on their own, so a decision does not have to wait for a specialist. Say or type what you are working on, and it routes you to the right task, grounded in TOPdesk's own product documentation. Twenty tasks across the research lifecycle, built as configuration rather than code.
View Case StudyAn internal tool, built and in use, putting six fragmented research and customer data sources behind a single plain-language query — qual and quant agents run in parallel, then a third synthesises findings shaped to the asker's role, so research data flows into product and AI strategy decisions.
View Case StudyInterviews and a discovery workshop with 7 stakeholders on why feature releases were inconsistent at TOPdesk, clustered into four categories of finding and turned into recommendations for product management, the UX guild, user guidance and marketing.
View Case StudyDoctoral and postdoctoral work in virtual reality.
Research exploring motivation patterns and behavioural differences in VR dance platforms, uncovering insights on confidence, engagement, and social connection.
View Case StudySix peer-reviewed papers on VR locomotion and interaction, plus the VR dance work published in The Visual Computer.