Heejin Yang
Frontend Developer
A frontend developer who came by way of cloud engineering and full-stack work.
I am the sole frontend developer at an AI brand-analytics SaaS. I own the frontend of seven web products — the analytics dashboard, internal back office, leaderboard, landing site, blog platform, company site, and an internal operations system. Four of them I built from zero, from requirements through deployment.
I work across the whole path: type definitions, API integration, the auth proxy, and deployment configuration. My strength is turning defects that no one can see by eye into automated verification gates, and building frontends that answer engines and AI agents can actually read.
The case studies here cover only the moments where a judgment call went one way rather than another. I tried to write down what I reverted and got wrong, not what went smoothly.
Career
- 2025.11 – present
FlipBFrontend
Sole frontend for an AI brand-analytics SaaS. Seven web products, four built from zero.
- 2025.02 – 2025.07
AconsoftDevOps
MSA consulting for a financial client. Frontend module federation design, approval-gated deployment.
- 2023.05 – 2024.05
HubertechFull Stack
Kubernetes-as-a-Service platform. Introduced MSW so frontend work no longer waited on the backend.
- 2022.05 – 2023.04
RabbitCloud Engineer
Open cloud platform. Automated security auditing, CCE and CVE vulnerability analysis.
Other work
Internal ITSM & groupware
Planned an in-house replacement for a collaboration tool being sunset. Wrote 17 specification documents and fixed the first release at 89 of 130 features — while explicitly listing the 46 we would not build. Implemented 17 screens for step one.
Product landing site
Built the lead funnel and the SEO/GEO surface. Removed duplicate canonical tags per route and shipped llms.txt for AI crawlers in both languages.
Multi-tenant blog platform
One codebase serving an independent blog per customer. Tenant branding is injected as CSS variables, so there is no per-tenant branching in code.
AI agent development environment
Configured coding-agent workspaces per repository — role-specific subagents, hooks, and skills. Rules are written as runnable check commands rather than prose to be interpreted, 7 to 13 per repository.