Masterful CV
An independent CV product for editing, previewing, exporting, and tailoring application materials.
Problem
Most CV tools split the work across too many places: one app for writing, another for exporting, a spreadsheet for applications, and an AI chat window for tailoring. That makes the important artifact, the CV itself, feel secondary to the tooling around it.
Approach
Masterful CV is built around a document-led loop: import, edit, preview, export, and track. Anonymous users can start locally, parse an existing PDF or DOCX CV, work with a live preview, and export a useful PDF without committing to an account. Signed-in users get a cloud-backed workspace for profile and job data, application-specific artifacts, onboarding, usage visibility, analytics, admin views, and AI generation flows for tailored CVs, cover letters, profile suggestions, and job-description guidance.
The product architecture treats profile data as the canonical source for document generation. Import flows expose what was read, what was mapped, and where parser or model limits may affect the result. Account and planned monetisation flows are designed as part of the product rather than bolted on: WorkOS handles identity and session recovery, Stripe supports credit and entitlement flows, and the usage flow makes AI costs visible before users spend credits.
Outside the AI paths, the product includes public marketing, product information, contact, FAQ, and legal pages; login and session-expired recovery; dashboard, profile, jobs, planned billing, usage, checkout, and admin surfaces; cloud sync with conflict recovery; locale-prefixed routing; privacy warnings; application package import/export; and print routes for CVs and cover letters.
The hardest product work is refining LLM workflows so generated output becomes useful, inspectable, and clear enough for users to review before relying on it. The generation pipeline uses schema-validated step artifacts, PII redaction, visibility filtering, retrieval-bounded evidence, allowed evidence sets, cover-letter alignment maps, multi-draft selection, repair and polish passes, regression checks, warnings, rate limits, abuse controls, partial-run recovery, generation history, and credit-aware model calls.
I originated the review/revision approach and refined it with agent assistance. I also orchestrated a 74-locale internationalisation effort using GPT, Claude, Gemini, and directed AI agents for locale quality checks, terminology parity, untranslated-string reporting, and route/locale smoke coverage.
I have designed and implemented the product end-to-end, while retaining product, architecture, verification, integration, and release decisions.
Outcome
Masterful CV takes a visitor through the main product path:
- starting from the public site
- editing a local profile
- previewing a CV as a document
- exporting a useful PDF
- signing into workspace routes for jobs, planned billing, usage, legal pages, and admin views
The credit-based, AI-assisted paths follow the same document-led principle as the local editor.
Release evidence in the repo includes a product-surface contract, security audit status, verification matrix, data-loss prevention matrix, cross-browser Playwright coverage, accessibility smoke checks, locale integrity checks, billing canon checks, and browser-flow screenshot manifests.