Full-Stack Product Engineer · AI-Native Builder
I turn messy requirements into working web products: data-heavy dashboards, admin panels, API-backed workflows, e-commerce systems and AI-powered tools. I work framework-agnostic and choose the stack that fits the product — the framework is not the product, the result is.
Live demos
A mix of full-stack products and frontend-heavy experiences you can click through. Built across different stacks on purpose.
Problem. Browser games often look like tech demos but do not feel replayable after the first click.
Built. A deterministic comet-racing arcade demo with 10 procedural levels, rival bots, biomes, hazards, jump/nitro controls, personal bests and mobile support.
Problem. Action demos need immediate game feel, clear feedback and enough systems to be worth replaying.
Built. A PixiJS objective-survivors game where light is both health and reach: charge pyre beacons, survive enemy pressure, unlock upgrades and open the Dawn Gate.
Problem. Agent workflows are hard to trust when you cannot see tools, retries, cost, trace history or approval gates.
Built. A deterministic execution console: node canvas, replay timeline, trace waterfall, token/cost HUD, command palette and human-in-the-loop approval editing.
Problem. Logistics teams need live map context, incidents, SLA drift and vehicle details in one operational view.
Built. A map-first control room over a synthetic fleet day: animated trips, replay scrubber, incident feed, depot filters, KPI strip and vehicle drill-down drawer.
Problem. Most portfolios show static screens; realtime graphics need performance, motion craft and careful browser engineering.
Built. An audio-reactive WebGL particle instrument with GPGPU simulation, Web Audio analysis, responsive controls, reduced-motion mode and cinematic interactions.
Problem. Premium storefronts need more than product cards: configuration, cart logic, checkout states and polish.
Built. A frontend-only e-commerce experience with animated product configurator, filters, cart drawer, mock checkout, order timeline and persisted saved builds.
Problem. E-commerce operations scattered across orders, inventory, returns and channels.
Built. A dense ops analytics console — KPIs, cross-filtering charts, and orders / inventory / customers / returns / insights with drawers, modals and bulk actions.
Problem. Outpatient intake & triage operations run by hand on spreadsheets.
Built. A queue + triage-review workflow with patients, appointments, follow-ups, operational reports and an audit log.
Problem. Personal finance data with no clear cash-flow, budget or category insight.
Built. Transactions, analytics, CSV import, budgets, subscriptions, savings goals, reports — plus a keyless AI assistant with per-page prompts.
Problem. Sales teams drowning in spreadsheets and manual order math.
Built. An order-accounting module: clients, inventory, server-computed orders, analytics, reports, client drill-downs and an activity log.
What's next
Categories I am actively building demos for next.
Conversational assistants with tools, memory, context and human-in-the-loop review.
In progressBuyer flows, seller/admin workflows, payments, catalog and order operations.
In progressETL-style cleanup and scheduled reporting pipelines over messy sources.
In progressMulti-view internal tools with auth, roles, dashboards and deployment.
In progressCapabilities
Frontend
Backend
Data & API
AI-assisted delivery
How I work
I turn messy requirements into a clickable, deployed demo quickly — then harden it.
Short, concrete updates. You always know what is done, what is next, and what is a trade-off.
I use AI agents to move faster, but every result gets architecture, testing and review before it ships.
Sensible structure, tests where they matter, Docker, and deployable from day one.