Good to see you :)
(run your cursor over my name. it's warm.)(tap my name. it's warm.)
i build products. with taste.
product person, designer, hands-on AI builder.
usually all three on the same day.
i ship real consumer AI: the experience people touch, the context behind it, the prompts, the evals, and the messy bits that only show up after launch.
i also build internal tools solo, first sketch to working system, fast. good UX isn't optional, it's part of the job.
i started in growth, obsessing over why people convert. then joined an API-first startup as its first PM. then shipped AI tools for thousands of creators.
today, i lead product, UX and AI at Outfit Formulas ↗, shipping alongside the founder in a team of fewer than ten.
i like being close to the whole loop. talk to users. find the real problem. design it. build v1. ship with the team. then see what actually happened.
an AI stylist grounded in stylist-authored logic, a user's wardrobe and their real constraints. i own the product direction across AI UX, context, recommendation logic and multimodal workflows.
the interesting work: making useful actions obvious, preserving styling intent through substitutions, and deciding when an expensive model call is worth it.
built ALI's human-evaluation workflow across 195K logged turns and 42K threads, with reproducible sampling and an internal review interface.
trace review, error analysis, annotation and readable context. i build the workflow around how a reviewer actually thinks, then use the findings to improve the product.
internal tools, end-to-end workflows and founder websites. i own the concept, information architecture, UI, workflow logic and implementation.
i run reusable skills and a deliberate UI/UX setup so AI gets context and the output stays consistent. then i check every detail myself.
i work across product events, subscription data and revenue systems to understand what's actually happening. i set up experiments in PostHog and trace surprising results back to user behaviour or the underlying data.
the question is usually more useful than another dashboard.
owned the AI scriptwriting copilot and designed Brainstorm, a multi-step creative workflow. deep user interviews plus behavioural data pushed us toward higher-value creation tools.
helped lift conversion 40%. Brainstorm pushed session completion up 35%.
wrangled API-heavy products across Shopify, WooCommerce, Amazon and Flipkart. redesigned onboarding and inventory UX, then built an analytics product 500+ customers used.
the onboarding and inventory revamp helped drive a 25% jump in platform adoption.
ran growth across $90K+ monthly e-commerce portfolios. behavioural deep-dives and experiments lifted conversion by up to 44%.
still my operating system: understand the behaviour first, then change things.
i live in AI as a builder and power user: prompts, context, structured outputs, model behaviour, evals. new tool drops? i'm testing it. the system around the model matters just as much.
most of my best calls came from a user call, a scrappy prototype, or a rabbit hole i swore i wasn't going down.
and if users don't care, i'd rather kill the bet than defend the roadmap.
i love ambitious products, fast-moving teams and people who sweat the details.
always up for a good problem. say hi.