In DevelopmentPublic project.
AI Portfolio Assistant
An AI-powered assistant that lets recruiters and hiring managers ask questions about experience, projects, stack, and role fit.
Role: Creator — Full-stack design, chat UX, API route, prompt engineering, and content curation.Duration: In Development
Next.jsTypeScriptOpenAI APIServerlessPrompting
Context
Static portfolios often fail to surface the depth behind projects. Recruiters scanning quickly miss architecture context, role fit signals, and full-stack ownership details.
Problem
A resume and project list cannot answer contextual questions like 'Is this person a fit for a NestJS + React role?' or 'What architecture work have they done?'
Ownership
- Chat UI with suggested question chips
- Server-side API route with prompt guardrails
- Curated portfolio knowledge base
- Rate limiting and safety rules
- Static demo mode when API key is unavailable
Architecture
Visitor interacts with chat UI → API route applies prompt and safety rules → curated portfolio knowledge informs LLM response → answer returned with optional case study links.
flowchart TD Visitor[Recruiter / Visitor] --> ChatUI[Portfolio Chat UI] ChatUI --> API[AI Assistant API Route] API --> Guardrails[Prompt + Safety Rules] Guardrails --> Knowledge[Curated Portfolio Knowledge] Knowledge --> Model[OpenAI / LLM Provider] Model --> API API --> ChatUI
Key Technical Decisions
| Decision | Why | Trade-off | Result |
|---|---|---|---|
| Curated knowledge over open retrieval | Prevent hallucination of private or unverified details | Less flexible than full RAG pipeline | Trustworthy, portfolio-accurate responses |
| Static demo fallback | Portfolio must work without API keys in all environments | Reduced interactivity in demo mode | Always-functional portfolio experience |
| Suggested question chips | Guide recruiters to high-value questions quickly | Less open-ended discovery | Faster time-to-insight for hiring managers |
Implementation
- Built floating AI orb with expandable glass chat panel
- Created suggested questions for role fit, stack, and case studies
- Implemented API route with environment variable for API key
- Added prompt rules to never expose secrets or confidential details
- Designed static demo mode with pre-curated responses
Results
- Turns the portfolio into an interactive interview preparation surface
- Demonstrates AI product thinking alongside engineering craft
- Differentiates from standard static developer portfolios
Reflection
- Guardrails matter more than model choice for a portfolio assistant.
- Linking answers to case studies keeps the experience credible.
- Next: recruiter email capture and usage analytics with privacy respect.