Selected Work

Most dating apps are built around swiping. We wanted to start somewhere else: getting people into the same room. findnaseeb is built around real events instead of profiles. I co-founded it with two other people (one engineer, one product/UX designer) and built the app.

• What I built: the web app, the APIs, the data model, and the AI step that shortlists who to invite to each event.
• The AI part: it looks at profiles and some context (location, who's new, who's been before, group balance) and drafts a guest list. We always review that by hand before anything goes out.
• A tradeoff I'd defend: we're a small team, so I kept the stack deliberately boring. Next.js and SQLite on a single box. It's easy to run and easy to reason about, and we'll move off SQLite when concurrency actually forces us to.
• Trust & safety: members can block anyone, and blocks come straight to us. We haven't needed heavy automated moderation. Meeting in person tends to self-select for people who behave.
• Where it's at: launched early 2026. About 200+ members in Calgary, all from organic Instagram and TikTok. Waitlists open in Toronto, Vancouver, and Chicago.

AMA's agent POS was an aging web app: bloated, buggy, and slow to change. When I joined we started rebuilding it. I was one of the first to lead the spike on micro-frontends and built the proof of concept we used to break the monolith apart. Once it was decoupled into MFEs, the app got faster and easier for agents to use, and it moved membership and product sales.

• What stuck around: I wrote the first template MFE repo, which other teams have reused many times, and a shared helper library for the functions every MFE needs. Together they cut a lot of duplicated code.
• Recent work: a new "Community" membership tier. It's a lower-cost monthly plan where members pick the benefits they want and skip the ones they don't.
• Faster to build and ship: spinning up a new app used to take days of boilerplate. The template and shared tooling brought that to about a day (roughly 60% less), and parallelizing the test suite made deploys about 65% faster.
• Better for agents: smoother transaction flows let agents get through about 40% more per shift, and Playwright catching regressions before release dropped production bugs about 30%.
• The hard part: micro-frontends aren't free. Shared dependencies drift across teams and get painful fast, so a lot of the work is keeping those shared packages stable so nobody gets blocked.
• Beyond code: I mentor new interns and take scrum roles on the team.

A marketplace that connects venues with casual staff. It runs payments between both sides and has a web admin panel.

• What I built: the backend. Job routing, payments between venues and workers, and the tools the ops team used every day.
• The hard part: marketplace payments. Money moving between two sides means edge cases everywhere: refunds, no-shows, disputes. Most of the work went into the unhappy paths.
• Stack: Node.js, PostgreSQL, AWS, Stripe.