Portfolio

Senior/Staff product engineer for AI-powered consumer products and platform architecture: from first architecture to an 8M+-user rollout.

I help companies and product teams turn ideas into AI-powered applications that scale: consumer web and mobile, LLM features with real evaluation and guardrails, payments and third-party integrations, and platforms where reliability matters.

As a hands-on technical lead, I set technical direction, own initiatives end to end, and partner with design, product, legal, and privacy. I have led UI architecture and a staged rollout to more than 8M users, and I now direct many AI coding agents in parallel, with the scope, standards and verification that keep fast work correct.

If you are looking to build, modernize, or add AI to a product, I bring 14+ years across large tech companies, finance, and startups, plus two published Node.js books and a long track record of mentoring engineers.

See recent projectsJourneyDoc, Bilt Base, Phloto and StatusGo, live and free to try
  • 10engineers mentored at Meta and Chelsea Piers
  • 8M+users reached by the Meta AI on Quest rollout
  • $100M+incremental revenue from Messenger ads work
  • 2published books (Manning, O'Reilly)

Recent projects

Owned end to end, from first idea to the App Store, and live where you can try them. Alongside them I run Hacky Apps, where I lead distributed engineers and contractors across client projects.

Toolbox

The languages, frameworks, and platforms I work with, grouped by area. Dots show depth: three means I reach for it daily; one means I have shipped with it and can get productive fast.

Showing 55 of 55 tools.

AI & ML

LLM features with evaluation and guardrails, agent workflows, and the ML work that came before.

  • LLM feature integrationDaily driversince 2024Meta AI on Quest · Marlow
  • Agent workflows & tool useDaily driversince 2026Marlow
  • Evaluation, telemetry & guardrailsDaily driversince 2024Meta AI on Quest · Marlow
  • Coding agents & MCP toolingDaily driversince 2025Daily workflow · This site
  • Applied MLShipped in productionsince 2019Meta Messenger · Bloomberg
  • scikit-learnWorking knowledgesince 2018Bloomberg
  • JupyterShipped in productionsince 2017Bloomberg

Languages

Typed JavaScript first; server and mobile languages as the surface demands.

  • TypeScriptDaily driversince 2018Chelsea Piers · Meta · Soon Dating
  • JavaScriptDaily driversince 2012Everywhere · Manning book
  • PHP / HackShipped in productionsince 2021Meta
  • PythonShipped in productionsince 2015Bloomberg · Meta
  • JavaShipped in productionsince 2012Android · Bloomberg
  • KotlinShipped in productionsince 2022Meta AI on Quest
  • RubyShipped in productionsince 2012Bloomberg · NYCDA

Web & design systems

Component-driven UI that is accessible, testable, and fast to extend.

  • ReactDaily driversince 2015Chelsea Piers · Meta · Hacky Apps
  • Next.jsDaily driversince 2018Chelsea Piers
  • VueDaily driversince 2017Bloomberg · Hacky Apps · This site
  • NuxtShipped in productionsince 2018Hacky Apps · This site
  • ReduxShipped in productionsince 2016React Native apps
  • D3.jsShipped in productionsince 2017Bloomberg
  • HTML & CSSDaily driversince 2012Everywhere
  • SassShipped in productionsince 2014This site · Rails apps
  • VitestShipped in productionsince 2022Chelsea Piers
  • Design systemsDaily driversince 2017Chelsea Piers · Bloomberg · Meta

Mobile

Cross-platform and native, through to app-store release.

  • React NativeDaily driversince 2020Soon Dating
  • AndroidShipped in productionsince 2012Meta · UIU-JVP
  • Jetpack ComposeShipped in productionsince 2022Meta AI on Quest
  • iOS & TestFlightShipped in productionsince 2020Soon Dating
  • FastlaneShipped in productionsince 2023Release pipelines
  • Push notificationsShipped in productionsince 2023Soon Dating

Backend & APIs

Service boundaries and contracts that stay clean as products grow.

  • Node.jsDaily driversince 2013Manning book · Meta · Soon Dating
  • ExpressDaily driversince 2014Manning book · Hacky Apps
  • GraphQLDaily driversince 2019Chelsea Piers · Meta
  • REST APIsDaily driversince 2013Everywhere
  • Ruby on RailsShipped in productionsince 2011Bloomberg · NYCDA · Hacky Apps
  • KafkaWorking knowledgesince 2019Bloomberg
  • DockerShipped in productionsince 2018Bloomberg
  • Queues & messagingShipped in productionsince 2021Meta Messenger

Cloud & data

Serverless backends, relational and document stores.

  • AWSShipped in productionsince 2016Soon Dating · Side projects
  • LambdaShipped in productionsince 2020Soon Dating
  • DynamoDBShipped in productionsince 2023Soon Dating
  • PostgreSQLShipped in productionsince 2014Hacky Apps · NYCDA
  • MongoDBShipped in productionsince 2015Manning book

Payments, integrations & observability

Third-party systems wired in without fragile glue code, and visible in production.

  • PlaidShipped in productionsince 2025Chelsea Piers (ACH)
  • AdyenShipped in productionsince 2025Chelsea Piers (cards)
  • StripeShipped in productionsince 2016NYCDA
  • Feature gates & staged rolloutsDaily driversince 2021Meta AI on Quest · Messenger
  • New RelicWorking knowledgesince 2016Bloomberg
  • SplunkWorking knowledgesince 2017Bloomberg

Tooling & practice

How the work stays shippable: source control, CI, design collaboration, and teaching.

  • Git & GitHubDaily driversince 2012Everywhere
  • GitHub ActionsShipped in productionsince 2020Chelsea Piers · This site
  • Jira & agileShipped in productionsince 2014Bloomberg · Meta
  • FigmaShipped in productionsince 2016Design handoff
  • Technical writingDaily driversince 2016Manning · O'Reilly · Articles
  • Mentorship & teachingDaily driversince 2013NYCDA · Meta · Chelsea Piers

What I bring

Seven areas of hands-on impact: what you can hire me to own, the tools I use, and proof from shipping.

  • Staff scope at AI speed

    Staff engineering with AI coding agents

    Setting scope, standards and verification for many parallel AI workstreams, so a fast-moving, global product lands in days and stays correct.

    • Coding agents & MCP tooling
    • Agent workflows & tool use
    • TypeScript
    • Next.js
    • MongoDB
    • GitHub Actions
    Selected experience 3
    • Led product and engineering for Mecka AI's partner and admin apps: took the partners app from a sign-in screen to a working product on production data in one week, for partner organisations in many countries, in five languages.
    • Ran about forty parallel AI coding sessions, each owning one pull request through tests, automated review and merge: 88 pull requests merged in four days, and the partners test suite grew from 56 to about 1,200.
    • Held a correctness bar where pay was involved: statements reconciled to the payment system to the cent, business sign-up proven on production against the internal tool, and a three-pass access audit that found and fixed five over-exposures.

  • AI that ships

    AI products & agentic systems

    LLM features people actually use: architecture, rollout, evaluation, and guardrails, not demos.

    • LLM feature integration
    • Agent workflows & tool use
    • Evaluation, telemetry & guardrails
    • Coding agents & MCP tooling
    • Python
    • GraphQL
    Selected experience 3
    • Set technical direction and led end-to-end UI architecture for the Meta AI Assistant on Quest: 42% engagement lift within 30 days of launch, then a staged 800K → 8M+ user rollout with telemetry-driven rollback.
    • Build LLM-driven agent workflows at Marlow that draft and file CMS-compliant nursing documentation into PointClickCare, with tool-use integration, prompt and output evaluation, and validation guardrails.
    • Earlier ML product work: heuristics-to-ML transition for Suggested-As-You-Type across Messenger clients, and NLP entity extraction over 100K+ agreements at Bloomberg.

  • Lead the whole arc

    Leadership, mentorship & cross-functional delivery

    Tech lead at Meta and Bloomberg: setting technical direction, directing work across seven teams, mentoring 10 engineers, and hiring senior talent.

    • Mentorship & teaching
    • Technical writing
    • Jira & agile
    • Figma
    Selected experience 4
    • Project and server tech lead across Messenger initiatives, including in-Messenger ad creation ($100M+ incremental revenue) and org-wide privacy remediation across seven teams.
    • Onboarded and mentored 6 engineers at Meta and 4 at Chelsea Piers; led external interviews contributing to 5 senior hires; ran weekly cross-team UI/UX reviews.
    • Set the ground rules for a week of fast parallel work at Mecka AI: never edit the system that pays people today, a frozen statement is the truth for a closed period, and access is enforced only on the server.
    • Author and educator: two Node.js books (Manning, O'Reilly), workshops, and curricula used in industry and universities.

  • Clarity at the boundary

    Architecture & APIs

    Service boundaries, contracts, and data flows that stay maintainable as products grow.

    • GraphQL
    • REST APIs
    • Node.js
    • PHP / Hack
    • Java
    Selected experience 4
    • GraphQL and REST design for membership, identity matching, and sync with legacy and third-party systems at Chelsea Piers; multi-step onboarding and account flows in production.
    • High-availability MetaAI APIs for Quest and the search integration spanning Quest and Smart Glasses; clear ownership of schema, versioning, and cross-system consistency.
    • Core messaging-layer changes at Messenger for 100M+ business accounts monthly and 1B+ user accounts, directing client rollouts across iOS, Android, and Web.
    • Two apps on one set of shared packages at Mecka AI, so partner and admin views compute the same numbers, with analytics reads scoped, bounded and cached per request to keep query cost down.

  • Product-grade UI

    Web, design systems & UX engineering

    Frontends that are accessible, consistent, and fast to extend, not one-off screens.

    • React
    • Next.js
    • Vue
    • TypeScript
    • Vitest
    • Design systems
    Selected experience 3
    • Reusable, accessible component library (forms, modals, steppers) and resilient multi-step signup with persisted state and cross-session recovery at Chelsea Piers.
    • Large-scale web at Bloomberg: the org's first standalone Vue application, Rails integration, and performance work for real user workloads.
    • Vitest, GitHub Actions CI, and Lighthouse audits so UI changes stay safe to ship.

  • Ship on every surface

    Mobile: native & cross-platform

    Shipping on iOS and Android, from React Native to native stacks and app-store delivery.

    • React Native
    • Android
    • Kotlin
    • iOS & TestFlight
    • Fastlane
    • Redux
    Selected experience 3
    • Led Android 14 migration and Jetpack Compose adoption across core Quest surfaces, reducing UI-transition crashes by 30%.
    • React Native at Soon Dating: iOS features end to end through App Store delivery, matching-algorithm iteration (+60% post-engagement retention), TestFlight and Fastlane pipelines, push notifications.
    • Cross-org coordination for mobile + server changes so releases stay predictable.

  • Connect what matters

    Integrations, payments & reliability

    Connecting your product to banks, card networks, providers, and internal systems without fragile glue code.

    • Plaid
    • Adyen
    • AWS
    • Feature gates & staged rollouts
    • GraphQL
    Selected experience 4
    • Payments: Plaid ACH and Adyen cards at Chelsea Piers, reducing payment-related onboarding failures by 28%; Stripe billing with installment logic earlier.
    • Staged rollouts, feature gating, and telemetry so launches and incidents are visible and controllable, from Quest device config to Messenger experiments.
    • Traced payment-run errors at Mecka AI to their causes (late-processed recordings never paid, quota deductions never refunded), shipped the fixes, and wrote a guide for Finance on which figures to trust.
    • Defensive client logic, async hardening, and monitoring patterns learned on high-scale consumer surfaces.