I build AI apps.

Paste a job listing and practice the interview out loud. It researches the company, predicts the kind of interview you’ll get, asks real follow-ups, then scores your answers and tells you what to fix. Built for students who don’t have someone in the industry to call before a big interview.

Scan a Pokémon card and get an instant estimated grade and value across PSA, CGC, TAG and Beckett. It also organizes the rest of your collection: virtual binders, set tracking, and live market prices so you always know what everything’s worth.

Companies run hundreds of security rules on their Linux servers and nobody really knows which ones work. I went through 754 public ones and not a single one comes with real numbers. So I’m building a way to test them: run the attack, see if the rule catches it, and if it doesn’t, figure out why. It’s already caught a hole in a rule people use in production.

After an outage, the team writes a postmortem, promises to add a check so it never happens again, and then nobody has time to build it. This builds it for them. It reads the postmortem, writes the check, and tests it against the actual code that broke and the code that fixed it. If it can’t prove the check works, it says so instead of guessing.