Show HN: A client-side adaptive cognitive test measuring micro-hesitations

Hi HN!
I’m not a software engineer by trade. My background is in behavioral science and human-computer interaction, but I wanted to solve a specific problem I found when applying for jobs which is the logical test. They are booooring, easy to cheat on if you really want to and provides very limited information about people. I built this prototype from scratch sing AI to bring my idea to life.
So now I've been working on this side project for a Cognitive Adaption Engine (CAT/KAT in swedish) for about three weeks. Many pre-employment screening tests i've done use a static item bank and are very liberal with the time allowed on each question, making it easy for Vision LLMs or bots to beat.
I wanted to see if I could build a fast, client-side engine (JS/SVG) that combats this problem by measuring certain biometric cognitive signatures like "switch cost" (how long it takes for your brain to re-adjust when logical rules change) and "post error-slowing".
To go beyond the simple grading of correct/incorrect, this test dynamically gives you waves of patterns/challenges through a local item bank based on your millisecond response times. If you use an LLM, the constant latencies gets…