I take enterprise SaaS products from requirement to production at regulated, industrial clients. 3+ years shipping B2B software, 4 years before that building supply chain operations from zero. I like problems with a compliance constraint, a messy client requirement, and a real deadline.
Nearly four years at Neosap Global — from associate PM authoring backlog to owning the roadmap for multiple enterprise accounts and leading an on-premise deployment end to end.
I also build. This is a grounded AI assistant I designed, built and deployed myself, end to end.
A grounded AI assistant for automotive showroom sales consultants. Answers spec, comparison and objection questions from a curated document corpus — and refuses plainly when the corpus can't support an answer, instead of guessing.
Three PM exercises from a 3-month fellowship — problem framing, RICE prioritization, GTM, and 0→1 concept design. Each links to the full write-up.
82% of users say they care about sustainability; only 11% actually opt in to eco-packaging or no-cutlery. Framed the problem end to end — persona interviews, survey data, RICE-prioritized feature set — and designed "My Green Defaults," a save-once preference layer with checkout nudges and a gamified Green Meter, plus a 3-phase GTM from 5% pilot to national rollout.
Segmented Spotify's free tier, chose "Casual Free Users" as the highest-leverage cohort, and sized the revenue opportunity: +10 min/day of listening ≈ $55M/yr in ad revenue and $43M/yr in incremental Premium revenue, backed by 5 user interviews on ad fatigue and discovery gaps.
Wireframes and flow for an AI-assisted feature concept on Bumble, exploring how an AI assistant could support the dating experience.
Training regularly and eating on a plan isn't separate from how I work — it's the same habit of showing up and tracking the numbers, applied somewhere other than a sprint board.
I follow model and product releases as they land, not on a lag — it's how I ended up building Showroom Copilot instead of just reading about RAG.