Medical and Clinical Analysis Products
Medical and Clinical Analysis Screen Development
I work on screens, accessibility and performance improvements, development environments, and backend tasks for medical and clinical analysis products.
- I developed input, execution, result, and report screens plus ECharts and Plotly comparison charts so users could move from condition setup through result review across long-running analyses.
- To let users continue when the viewport changed or they returned after an error, I proposed responsive, error, and re-entry behavior that was missing from the plan and defined and applied it with product, design, and backend teammates.
- To reduce repetitive input and mouse movement on input-heavy screens, I added keyboard tab navigation and import features.
- To reduce image requests and the scope of UI changes, I replaced image-based UI with shared components and added a build task that converts static images into viewport-specific WebP assets. Calculated image transfer size fell by about 95%, while a separate Lighthouse comparison reduced LCP from 1.3 to 1.0 seconds and Speed Index from 3.4 to 0.7 seconds.
- To avoid requesting the same disease list on every screen, I replaced direct requests with a shared TanStack Query hook and managed cache and refetch rules in one place.
AI Development Tooling Experience
- To keep AI from interpreting medical and clinical terminology and screen behavior on its own, I stored project instructions, task-specific Skills, execution Hooks, and PR rules in the repositories for Codex and Claude to reference.
- To make domain material easier for non-engineering roles to understand, I converted planning and history documents to Markdown and built and deployed an MCP server backed by a local LLM Wiki searchable from Codex and Claude.
Development Environment and Backend Operations
Tech Stack
Next.js
React
TypeScript
Tailwind CSS
Radix UI
TanStack Query
Zustand
React Hook Form
ECharts
Plotly
Vitest
Playwright
axe-core
GitHub Actions
Husky
Docker
Python
Django
Django REST Framework
PostgreSQL
Codex
Claude
MCP
- I unified the structure and deployment flow of four frontend projects split between Vite and Next.js Pages Router, then updated the existing code to a shared coding convention.
- I led the move from manual Docker deployment to GitHub Actions and made the workflow leave an Actions Summary and logs when backend HTTP responses or container status were abnormal after deployment.
- I connected typecheck, test, and build to Husky pre-commit and pre-push so type, test, and build failures could not reach remote branches.
- To trace service errors by product, environment, and request, I separated logs from business databases across three backends and stored them by environment, service, and request ID. In Admin, I added log-query and frontend error-reporting APIs and mapped frontend error types to server logs.
