Microsoft: UXR for Python, Developer, and AI Experiences

Quick Pulse studies are one of Microsoft’s continuous-learning mechanisms: a lightweight study to engage customers weekly—creating a reliable pulse on product health and opportunity. I initially ran QPs solo end-to-end; as scope expanded, I onboarded and managed a contractor while staying accountable for quality, rigor, and synthesis. I partner with PM, Engineering, and Design to set learning goals, pressure-test hypotheses, and turn evidence into roadmap moves across VS Code, Azure/Notebooks in VS Code, Documentation, Azure, and more.

Scope & responsibilities

• Own research planning, protocols, and success criteria; ensure methodological fit and signal quality.

• Facilitate sessions, drive rapid synthesis, and align cross-functional stakeholders on decisions.

• Manage recruitment/screeners and lab/remote setups; standardize templates/playbooks; oversee contractor execution.

• Make sense of quantitative signals in light of qualitative feedback.

Methods applied
Moderated interviews • Rapid usability (task-based, think-aloud) • Observation/behavioral coding • Contextual inquiry • SUS + short-form surveys • Heuristic/cognitive walkthroughs • Mockup/prototype reviews • Concept Value Tests (Kano/CVT) • Flow evaluations • Telemetry triangulation.

Impact

• Direct product impact: shipped new features, refined/renamed patterns, and tightened workflows based on week-over-week evidence.

• Connected “disconnected” team study sessions into cross-study patterns and emergent cohorts—unlocking clear prioritization and net-new opportunities.

• Shortened the learn→build loop and de-risked releases via a repeatable, closed-loop research cadence adopted by partner teams.

Born out of Quick Pulse insights, I connected patterns across otherwise unrelated studies to expose cohort-level friction in VS Code and throughout our Azure E2E journey. Leadership stood up a dedicated E2E team as a result; I owned the Python and Node tracks while continuing to run the Quick Pulse pipeline.
Resulting in the development of a cross-product/experience adoption funnel that teams now use to prioritize, sense-make, and track experience.

Scope & responsibilities

• Own research for Python/Node E2E across Azure services; set learning goals with PM, Eng, and Design.

• Facilitate sessions and workshops; drive rapid synthesis and decision alignment.

• Manage recruitment/screeners across cohorts (new-to-cloud, migrating, advanced).

• Triangulate telemetry, surveys, and product usage with qual findings.

• Communicate insights and implications to leadership.

Methods applied
Moderated interviews • Contextual inquiry • Multi-day/diary studies • Participatory co-design • Task-based usability • Workflow observation/behavioral coding • Service blueprinting + journey/funnel mapping

Impact

• Direct product changes to onboarding, environment setup, auth, and deployment flows for Python/Node; clearer first-run and tighter workflows.

• Standardized “E2E Adoption Funnel” used by partner teams to prioritize and make sense of data.

• Converted scattered signals into org-wide guardrails and playbooks—raising quality bars and accelerating time-to-decision.

Multi-year focus on the DS notebook experience. I drove definition, validation, and iteration for core features—history diffing, Gather (replicating cell results), line-by-line debugging—and led the research track behind Data Wrangler to reduce data-cleaning pain.

I facilitated Python team workshops, resulting in 40+ concepts, stood up a weekly validation cadence, and introduced a lightweight spec to align language and fidelity (works with a basic description, a user flow, or a working prototype).

When AI hit, I adapted priority concepts for the new paradigm and evolving customer needs.

Scope & responsibilities

• Own DS research across VS Code (Python/Jupyter) and Azure Notebooks; partner with PM, Eng, and Design on problem framing and decision criteria.

• Facilitate concept ideation workshops; drive backlog triage and sequencing; coach teams from idea → testable artifact → decision.

• Manage recruitment/screeners for varied DS personas; keep a steady learn–build loop with weekly testing.

• Triangulate qual with surveys/telemetry; maintain a concept portfolio and funnel for evidence-based prioritization.

• Operationalize a reusable, lightweight concept spec and playbooks for rapid research.

Methods applied
Mockup/prototype studies • Observational usability (task-based, think-aloud) • Contextual inquiry • Diary + remote ethnography • Concept Value Tests (Kano/CVT) • Heuristic/cognitive walkthroughs • Workflow analysis •

Impact

• Direct product changes: shipped history diffing, Gather, line-by-line debugging, and drove the research behind Data Wrangler—reducing copy-paste thrash and accelerating data prep.

• Established a reusable CVT framework and weekly test cadence that accelerated decision cycles and cut concept rework.

• Prioritized and matured a 40+ concepts; coached teams to move faster with clearer problem statements and acceptance criteria.

• Adapted the concept set for the AI era, preserving user value while aligning with new capabilities and guardrails.

• Improved notebook UX (state clarity, debugging flow, reproducibility) and reduced friction in the E2E data-cleaning pipeline.

When AI reshaped the roadmap, we led a research-intensive PM customer workshop series and turned the signal into a Customer Capability Model (CCM)—staged maturity with verifiable indicators. I owned end-to-end synthesis (reviewed every interview, built customer cards, codified patterns) and co-operationalized the model across teams to guide Copilot, AI Toolkit, and intelligent-app investments.
Discovery work on “AI playgrounds” fed directly into AI Toolkit concepts; as agent use cases emerged, I led a large-scale survey of agentic development to map practices, risks, and prioritization. In parallel, I stood up lean, repeatable research loops to get answers faster without diluting rigor.

Scope & responsibilities

• Co-define the CCM and maintain stage criteria; align PM, Eng, and Design on maturity targets and proof points.

• Run/oversee weekly AI Toolkit studies; shift cadence toward portfolio-level questions as scope scaled.

• Review sessions across studies, create customer cards, and synthesize cross-product patterns, pains, and opportunities.

• Partner with leaders on AI workshops; translate findings into frameworks, roadmaps.

• Drive discovery for Copilot experiences and LLM development workflows; keep a tight learn→decide loop.

• Lead an agentic-development survey; operationalize insights for sequencing and resourcing.

Methods applied
Participatory interviews • Mockup/prototype reviews • Concept Value Tests (Kano/CVT) • Task-based usability • Contextual inquiry • Remote ethnography • Short-form surveys/SUS • Large-scale surveys/segmentation • Workshop facilitation

Impact

• CCM adopted across teams to standardize language, track capability growth, and de-risk AI investments.

• Direct product impact: net-new AI Toolkit concepts and Copilot experience guardrails informed by real workflows.

• Exposed cross-study patterns (evaluation needs, versioning pain, orchestration gaps) that reshaped backlog priorities.

• Established a fast, repeatable CVT framework and lean research loops to accelerate decision velocity.

• Agentic-development survey created a shared map of the landscape—guiding roadmap, sequencing, and partner alignment.

• Converted fragmented signals into an operating system for AI work: principles, playbooks, and metrics teams now use to make consistent, inclusive, high-impact decisions.