RezumeScan — AI-Powered Resume Analyzer

The Challenge

Most job seekers don't realize their resume never reaches a human. Over 75% of resumes are filtered out by Applicant Tracking Systems before a recruiter ever opens them — not because the candidate isn't qualified, but because the resume doesn't match the right keywords and formatting the system expects.

Tailoring a resume for every application is tedious and mostly guesswork. Job seekers paste keywords they think matter, rewrite bullet points blindly, and hope for the best. There's no feedback loop — they never know why they didn't get a callback.

The tools that do exist are built for the other side of the table. Enterprise ATS platforms serve recruiters and HR teams, not the candidates being filtered out. Consumer-facing alternatives offer surface-level tips ("use action verbs") without analyzing the actual job description.

We saw an opportunity to build something better — a tool that gives job seekers the same intelligence that ATS systems use against them. And as a team that builds software products for a living, we wanted to prove we could ship our own.

Our Solution

We built RezumeScan as a full SaaS product from the ground up — authentication, billing, usage-based pricing, resume management, and the core AI analysis engine.

The product is simple to use: upload your resume, paste a job description, and get an instant analysis. RezumeScan scores your resume across three dimensions — keyword matching, expertise alignment, and overall fit — then provides specific, actionable feedback on what to improve. Not generic advice like "add more metrics," but targeted suggestions tied to what the job description actually asks for.

Making AI fast and accurate

The core technical challenge was making the AI analysis both meaningful and fast. Generic AI responses ("your resume looks good") aren't useful. But deep analysis that takes 30 seconds kills the user experience.

We integrated Google Gemini as our AI backbone and invested heavily in prompt engineering to get consistently structured, actionable output. The analysis breaks down exactly which keywords are present, which are missing, and how the candidate's experience maps to the job requirements — all returned in under two seconds.

Full SaaS from day one

RezumeScan launched with three pricing tiers — Free (5 scans/month), Basic ($12/month, 20 scans), and Pro ($30/month, unlimited) — powered by Stripe for billing and subscription management. Usage-based rate limiting was built into the architecture from the start, not bolted on after launch.

The entire product runs on a Next.js full-stack architecture with a Node.js API layer and PostgreSQL for data persistence. Using Next.js for both the frontend and API routes meant a single deployment target, which was critical for a two-person team moving fast.

We shipped a working MVP in three weeks, then iterated to full launch over the following months — adding resume management, detailed analysis views, and the tiered pricing model.

Key Technical Decisions

Single-stack deployment. We chose Next.js for both frontend and backend instead of splitting into separate services. For a two-person team, the reduced operational overhead was worth the trade-off. One repo, one deploy, one set of logs.

Model-agnostic AI layer. The Gemini integration sits behind an abstraction layer so we can swap AI providers without rewriting business logic. This protects against API pricing changes, rate limit issues, or better models becoming available.

Usage metering from day one. Rather than launching with unlimited access and adding limits later (which always frustrates users), we designed the tiered usage system into the billing architecture from the start. Every scan is tracked, limits are enforced gracefully, and upgrading is frictionless.