Almost Senior: product design of an AI job search app

Almost Senior is a new way to land a job: fresh roles every day, a CV adapted to each one without inventing anything, and every application tracked. An MVP, designed and built end to end. Below are the details that make it feel calm.

Onboarding starts with the file people already have. Parsing shows its progress field by field, so a minute of waiting reads as work being done, not a spinner. The result opens straight in the editor with a live preview.

The feed is a two-pane workspace: the list on the left, the job on the right, keyboard first. No tabs and no back button, so you can scan dozens of roles without losing your place.

A list that remembers. Opened jobs fade and get a 👀 on the logo, so what's new stands out without badges or counters.

Filters show the market before you choose. The salary histogram reflects the current list in your currency and period, and the button says how many jobs you'll get before you apply it.

My jobs is one table for the whole pipeline: CV, cover letter and status for each job. A sent application goes quiet with a 🤞, so the rows that still need you stand out.

15 CV templates, each with four palettes. The palette you pick goes straight into the PDF, free and without watermarks.

Templates are checked against 20 ATS parsing rules: one text flow in reading order, real text, standard section names. The typography sits on top, so a parser and a person read the same CV.

Tailoring is controlled, not magic. Pick a built-in mode or make your own on five-step scales; the app shows the nearest built-in mode and a before/after example. Companies, titles, dates and numbers never change.

An MVP, built to ship fast and stay cheap to run.

Web app: Nuxt 3, Vue and Tailwind.

Data and auth: Supabase, with each user's data kept private and secure.

AI: CV tailoring, cover letters and parsing run through OpenRouter, only on providers that don't store data.

PDF: CVs render to PDF on AWS Lambda.