Chiho Portfolio
A production portfolio that combines polished frontend delivery with structured content, explainable vacancy matching and operational engineering.

Business value
Frontend experience
Next.js, React, TypeScript, Styled Components and Motion deliver a responsive and accessible recruiter experience.
Backend and integrations
Node.js API routes, Python FastAPI, Contentful GraphQL and Supabase connect content, matching and protected data flows.
AI engineering
Gemini, Groq, embeddings, semantic search and validated structured output power explainable matching and the career assistant.
Quality and delivery
GitHub Actions runs builds, tests, SonarQube Cloud, CodeQL and Trivy before Vercel stages the exact approved revision. API and browser smoke tests then gate production, reducing release risk and shortening feedback cycles.
Overview
I designed and built this portfolio as a maintained product rather than a static showcase. It connects recruiter-facing case studies to structured evidence and makes the delivery setup visible and testable.
Because the portfolio includes public AI and contact features, I added layered security. Origin checks, input validation and rate limiting reduce abuse; sensitive data is masked before storage, credentials stay server-side, and security headers and hardened containers add further protection.
Security approach
Origin validation
Confirms that public API requests come from an allowed website, making cross-site abuse harder.
Schema validation
Checks input structure, type and length before processing, rejecting malformed or unexpected data.
Rate limiting
Limits requests per visitor and time window to reduce spam, AI cost abuse and service overload.
Sensitive-data masking
Redacts email addresses and phone numbers before conversation content is stored, while secret-like input is rejected.
Server-side credentials
Keeps model and Supabase service credentials out of browser code so visitors cannot read or reuse them.
Security headers
CSP, HSTS, frame protection and MIME-type protection reduce common browser attack paths.
Hardened containers
Runs the matching service as a non-root user in a minimal container, limiting the potential impact of a compromise.
The problem
A senior portfolio needs to prove more than visual quality. It should make technical decisions, delivery maturity and verified project evidence easy to inspect without overwhelming a recruiter.
My contribution
I built the responsive Next.js and TypeScript experience with React, Styled Components, Motion and accessibility-conscious interaction design. I implemented Node.js API routes and a private Python FastAPI matcher, integrated Contentful through GraphQL and use Supabase for protected data flows. I applied Gemini and Groq models, embeddings, semantic search and validated structured output to build explainable AI features. I created Docker Compose development and production environments with health checks and hardened containers. I also built the GitHub Actions CI/CD workflow with formatting, linting, type checking, Jest tests, Python matcher tests, evaluation and Docker integration checks before a gated Vercel deployment.
Tech stack
- Next.js
- React
- TypeScript
- Styled Components
- Motion
- Node.js
- Python FastAPI
- Contentful GraphQL
- Supabase
- Gemini
- Groq
- Embeddings and semantic search
- Zod and Pydantic
- Docker
- GitHub Actions CI/CD
- Vercel
- SonarQube Cloud
