AI PORTFOLIO PLATFORM
Dynamic portfolio management platform with AI résumé ingestion and database fallback.
Extracts employment dates, achievements, and technology lists directly from uploaded documents.
LLM is constrained to produce strictly validated TypeScript models with zero syntax errors.
Commits to Supabase PostgreSQL while updating local static JSON cache to ensure zero downtime.
Operational & Technical Friction
Updating technical portfolios requires tedious code redeployments or reliance on brittle third-party CMS services.
Systemized Implementation
Built a private full-stack management suite that uses generative AI to extract projects and experience directly from uploaded CV PDFs.
How It Operates in Production
Engineered with Next.js App Router, Supabase Auth, and OpenAI/Gemini structured outputs. Built an admin cockpit with real-time editing, draft previews, and automatic JSON synchronization.
Implemented strict Zod schema validation on LLM JSON output, error boundaries, and a transparent fallback provider ensuring continuous read availability.
Data Flow & Infrastructure
Uploaded resumes are processed by a server-side parser extracting raw text. The text is passed to an AI LLM with a strict JSON schema. Once validated against Zod models, the data is committed to PostgreSQL and published to the live portfolio.