For the past year I’ve been building JobSpace AI — an ATS CV checker focused on the UK job market.
That UK part matters more than I first expected.
A lot of CV tools seem to be built around US resume advice. But UK CVs are different. Two pages is normal. Photos are usually a bad idea. Dates are DD/MM/YYYY. Education often includes GCSEs, A-levels, degrees, professional qualifications, and sector-specific requirements.
The systems are different too. NHS Trac, Oleeo, Workday, Civil Service applications — they all have their own quirks.
The stack is fairly simple:
- Next.js 15 frontend
- Python/FastAPI backend
- Supabase for auth and storage
- GPT-4o for the free scan
- Better/tiered models for the paid reports
A few things that stood out from the data:
Around 35% of CVs score below our 75% pass threshold, even from people who clearly have good experience.
The biggest issue is usually not formatting. It’s keyword mismatch.
For example, someone might write “customer support” throughout their CV, but the job description says “customer service”. A human knows those are basically the same thing. An ATS may not.
NHS and Civil Service CVs are also much less forgiving than private sector CVs. The wording requirements are often very specific, and good candidates can lose points simply because their CV doesn’t mirror the role closely enough.
The most annoying technical problem has been PDF parsing.
Multi-column CVs, embedded tables, weird fonts, exports from Canva, and badly generated PDFs all fail in different ways. I ended up having to build a pre-processing layer to clean and normalise the CV structure before running the analysis.
The product itself is simple:
First scan is free, with no sign-up required. Users get a score and some example fixes. The full report, including all missing keywords and section-by-section suggestions, sits behind a paywall.
Happy to answer questions about building this, ATS systems, UK CVs, or the technical side.
jobspaceai.com/ats-resume-checker
Top comments (0)