CVHancer is our own product: an AI resume builder that tailors a CV to each job and checks it the way screening software does. Here is what we built, the technology behind it and the decisions that made it fast, affordable to run and trustworthy.
GuidesUpdated 12 October 20264 min read
Most candidates send the same resume to every job ad. Applicant tracking systems (ATS) and recruiters then filter it out because it misses the keywords of the role, uses a layout software can't read, or follows the wrong conventions for the country.
Generic AI writers make this worse: they invent achievements that the candidate then has to defend in an interview. We wanted a tool that adapts a resume to a job using only what is already true in it, and explains every suggestion.
Around the product sit the parts every SaaS needs: accounts, plans and AI credits, payments, transactional emails, an admin dashboard, analytics and error monitoring.
| Layer | Choice | Why |
|---|---|---|
| Web app | Nuxt 3, Vue 3, Tailwind CSS | Fast, server-rendered pages that rank on Google and feel like an app once loaded. |
| API | NestJS, MongoDB | A structured, typed backend with one module per domain: resumes, AI, billing, admin. |
| AI | LLMs through OpenRouter | One API for several models, so each task uses the best model for its cost. |
| PDF export | Headless Chrome (Puppeteer) | Pixel-accurate, text-based PDFs that ATS software can read. |
| Payments | Dodo Payments | A merchant of record that handles international sales tax. |
| Operations | S3 storage, Sentry, Docker | Reliable file storage, error alerts and repeatable deployments. |
1. The ATS audit uses no AI. The checks run in the browser on every keystroke, with fixed weights: a pass earns all the points, a warning half, a fail none. The score is instant, free to compute and always explainable, and checks that don't apply are left out rather than counted as passed.
2. AI is metered and monitored. Every AI action costs credits, each plan includes a monthly allowance, and daily AI spend is recorded. If usage spikes, the team gets an email alert before the bill does.
3. Privacy by design. The browser extension reads a job ad only on the tab the user clicks, and passes it in the URL fragment, which browsers never send to a server. Nothing goes to our API until the user acts.
4. One codebase everywhere. The website, the installable web app and the Android app (a Trusted Web Activity) share the same code, so every deployment updates all of them at once.
CVHancer's interface is available in English, French, Dutch and Arabic, including a full right-to-left layout for Arabic. The audit messages, country formats and templates are translated too, not just the menus. Building this in from the start costs far less than adding it later.
The first commit was made in early August 2026. About ten weeks later, CVHancer is live with free and paid plans, after more than 350 commits. Shipping a first usable version early and improving it with real user feedback mattered more than planning every feature upfront.
CVHancer uses the same approach we bring to client work: a solid SaaS foundation, AI used where it adds real value, and costs kept under control. If you are planning a SaaS or an AI-powered product, see our custom SaaS development and AI automation services, or try CVHancer(opens in a new tab) to see the result.
What it really costs to replace spreadsheets with a custom internal app in 2026, what drives the price, and how to tell when it pays for itself.
How much a custom CRM costs to build in 2026, when it beats HubSpot or Salesforce, and how to scope one without overspending.
How AI reads supplier invoices and sends them to your accounting software, what accuracy to expect, and what it costs a small business in 2026.
