I'm a full-stack dev who likes owning projects end-to-end. I co-founded BeJeweled, a live silver jewellery brand, and built the whole platform solo — React on the frontend, a Sanity Studio CMS so we can update 200+ products and pricing without touching code, and the full purchase flow (cart, Clerk auth, checkout), all shipped on Vercel with CI/CD running behind it.
Alongside that, I've done a good amount of LLM evaluation work — head-to-head comparisons across 10+ projects, writing the adversarial prompts and code-execution tasks that actually break models, and building grading criteria in containerized environments so the checks hold up. I did similar work at Ethara AI, evaluating outputs against structured rubrics and helping validate training data that fed into post-training.
