We turned complex, hardware-bound tokamak tools into an accessible, browser-based platform for students and researchers.
Fusion research tooling is powerful but hard to set up, scattered across local software and heavy infrastructure — a barrier for students and early-career researchers.
I was the sole Product/UX Designer in a small cross-functional team (PM, full-stack, back-end, mathematician).
Fusion research tooling is powerful but hard to set up, scattered across local software and heavy infrastructure — a barrier for students and early-career researchers.
• 70% of interviewees said existing tools are too complex or resource-heavy.
• 80% asked for a single platform that unifies simulation + data analysis.
• 65% struggled to collaborate due to fragmented toolchains.
Personas: a university professor (needs verified data & teaching-friendly UX) and a master’s student (needs hands-on experiments without steep setup).
• Democratize access — run simulations & analyze plasma data in the browser (no installs).
• Shorten time-to-insight — clear flows, sensible defaults, and ready-to-use visualization.
• Be lab-friendly — responsive UI, safe data handling, scalable components for future features.
• Workspace & File Manager: Upload and organize datasets (e.g., HDF5), open graphs, manage mappings.
• Guided Simulation Wizard: Step-by-step creation with pre-validated digital replicas of tokamaks and ML models; defaults reduce error risk.
• Mapping & Comparison: Side-by-side experimental vs simulated data; configurable graph panels for quick pattern reading.
• Responsive UI: Full web app + mobile access for on-the-go checks and learning.
• Design System: Tokens, components, states, and patterns for tables, graphs, forms; cuts delivery time and keeps quality consistent.
• +30% navigation efficiency (measured on task completion during testing).
• Zero-install access on web and mobile lowered the entry barrier for students.
• Platform adopted by students and researchers; foundation laid for collaboration features and future modules.
Designing for a niche scientific domain benefits from opinionated defaults and evidence-first visualization. Next steps: role-based permissions, real-time collaboration, and deeper notebook (Jupyter) workflows.