Corpus: Canonical Career Record
Generates ATS-parseable CV variants from a canonical master profile, where every data point retains its provenance and the AI does not hallucinate.
01 Problem and context
Maintaining multiple resume versions leads to diverging copies: updating a detail in one leaves it outdated in the others, and no one remembers where each figure came from. In addition, flashy layouts are often unreadable to applicant screening systems.
02 Solution
A single master profile from which variants are derived by reference rather than duplicated: updating a data point once corrects it everywhere. Every entry retains its provenance, and the model does not fill in what has not been recorded. The output is a single-column, true-text PDF readable by screening systems. It is currently in its foundational phase, not a finished product.
03 How it's built
Professional record system based on a canonical master profile from which ATS-parsable CV variants are derived. The variants reference the original profile to ensure every data point maintains its provenance and the AI never fabricates information. The project is built with Next.js 15, strict TypeScript, Tailwind v4, and Supabase.
04 Highlights
- Canonical master profile deriving CV variants via references to ensure complete data provenance.
- Single-column real-text PDF output formatted for ATS parsing with zero AI hallucination.
- Built with Next.js 15 App Router, strict TypeScript, Tailwind v4, and Supabase with RLS.
- Custom design system featuring color tokens and a WebGL aurora animation rendered on the auth route.