pubspec.yaml, Package.swift, and build.gradle.
There is something deeply satisfying about writing an abstraction that other developers actually adopt. Good SDK design is an underrated skill — you have to get the ergonomics right, on the first try, because every typo becomes someone else's stack trace at 3 a.m. I spend most of my time thinking about that.
I'm a mobile engineer with six years of experience and a particular obsession with building things other developers use. If it ships as a dependency in someone else's pubspec.yaml, Package.swift, or build.gradle— I'm interested.
“Cross‑platform without the excuses. Flutter for shared logic, Kotlin for Android-specific performance work, idiomatic Swift when iOS needs native behaviour. One solution does not fit everything.”
I build AI products, but not the “added a GPT call to a button” kind. The kind where the intelligence is the product — RAG pipelines that actually retrieve, vector stores that fit in your pocket, agents that know when to be quiet.
Open source is where I keep myself honest. If you've ever had a stranger leave a thoughtful review on your PR at midnight, you know why people do this. I try to be that stranger.
Outside of code: I love God, I love coffee, and I carry the irrational confidence that this time the build will pass on the first try.
Architect and maintain a cross-platform fintech SDK in native iOS (Swift), Android (Kotlin), Flutter and React Native — distributed to banking applications nationwide. I lead enterprise integrations and optimise the core financial services infrastructure end-to-end.
Developed an Android health platform integrating wearable BLE devices for bioresonance therapy. The work was almost entirely under the widgets — low-level Bluetooth optimisation, offline-first sync, and HIPAA-grade data pipelines for real-time biometrics.
Led end-to-end development of a stylist booking platform with a small team. Owned the Flutter codebase, the GraphQL surface, and the AWS Amplify infrastructure underneath it. Quiet release, real traction.
Promoted from Flutter Engineer to Lead. Built an automobile support platform from zero to 10,000+ users, and the Vendoor / Vendoor Rider logistics SDK — Android & iOS plugin, end-to-end. Pleasant work; the kind that makes you a better library author.
Spearheaded development of a ride-hailing platform with real-time tracking and a WebSocket-based dispatch system. First role where I really got to mentor — turns out that's the part of senior I like most.
100% native core for iOS & Android, with thin Flutter and React Native bindings. Tracks low-level telemetry continuously, flushes atomically at crash time. Designed to feel like having a black box on every device.
Mobile RAG: Google Gemini + a VectorDB for high-speed semantic search. Latency-budgeted so questions feel like conversation, not a request.
AI-powered wardrobe management PWA. Combines weather data, your closet, and a personal style model into outfit recommendations you actually wear.
100% native iOS / Android implementation under a small, ergonomic Dart surface. The kind of plugin where the README is shorter than the API, on purpose.
A mobile-first DSA gym with SM-2 spaced repetition that knows what's broken, a skill tree that shows what unlocks next, and five interaction modes so you stop grinding the same problem and start understanding it. Built offline-first — the full curriculum and review schedule sync to the device so you can study on a flight with no signal.
Improvements to the official Flutter analytics SDK. Quiet contribution work — the kind where you read a lot of CONTRIBUTING.md and learn someone else's architecture before you touch anything.
Contributions to Stripe's Flutter SDK. The standard for ergonomic payment libraries; a good place to sharpen the craft of 'does this API feel obvious'.
This site was designed and built in collaboration with an AI assistant — a deliberate choice, not a disclaimer. The composition, the type system, the marginalia, the rhythm of the sections: all iterated with the model as a peer in the room. The intent is mine; the execution was a conversation.
I work this way because AI is a tool I use professionally — for RAG pipelines, for code review, for the long boring middle of any project. It belongs on the portfolio that describes my work.
Nothing here is generated at you. There are no auto-rendered hero images, no purple-to-pink gradients, no oversized chat bubbles pretending to be a product. Every line of type was chosen, every margin set on purpose.