The languages, databases and tools I build with.

I get asked what my stack looks like fairly often. This is what I’ve actually shipped production systems with over the last decade or so — across founding-engineer work, startup platforms and large-scale infrastructure.

Languages & frameworks

  • Python (FastAPI)

    My default for backend services today. The Havenwise API platform runs on FastAPI, and it’s what I used to stand up UEFN AI’s infrastructure for autonomous agents in games. Before that, Django at Cargobase.

  • Go

    The workhorse for high-throughput services at Grab — invoicing, reconciliation pipelines and internal platforms handling millions of rows a month. Also what I standardised on for parts of Jelly’s backend.

  • Node.js & React

    The combination I’ve built the most product with: Championfy’s platform and admin portal, Ampverse’s AmpCreds marketplace, and a long tail of internal tooling.

  • Expo (React Native)

    How I ship mobile without a separate iOS and Android team. Built Championfy’s cross-platform apps and, more recently, Havenwise’s app from scratch to replace a no-code product.

  • Ruby on Rails

    Still one of the fastest ways to get a real application in front of users. Used heavily on internal platforms at Grab.

Data & infrastructure

  • Supabase & PostgreSQL

    Postgres with batteries included — auth, storage and row-level security out of the box. It’s the data layer behind Havenwise.

  • MySQL & MongoDB

    MySQL for most of the transactional systems I’ve built at Grab, Cargobase and Championfy; MongoDB (and DocumentDB) where the shape of the data kept moving.

  • Redis

    Caching, rate limiting, queues, leaderboards. The first thing I reach for when a database starts feeling the load.

  • Kafka

    For event streaming at scale — onboarding 10+ data streams into a central warehouse for reconciliation and fraud detection at Grab.

  • AWS

    ECS, SQS, Lambda, DocumentDB and friends. I migrated Championfy’s entire infrastructure from Google Cloud to AWS and cut thousands of dollars a month off the bill. AWS Certified Developer – Associate.

  • Google Cloud

    Cloud Functions, Cloud Run and the rest of the stack behind Havenwise, and where a lot of my early Google-era work lived.

  • Terraform

    Infrastructure that’s reviewable in a pull request. Non-negotiable once more than one person is touching the environment.

AI & developer tools

  • Claude

    Powers the conversational side of Havenwise AI, and it’s in my editor and terminal most of the day.

  • Cursor

    Where I write most code now. The gap between “I know what I want” and “it’s running” has never been shorter.

  • MCP

    Model Context Protocol is how I wire models to real systems — I built an MCP integration with Intercom’s Fin AI so it can diagnose heat pump faults directly against our platform.

  • CI/CD pipelines

    Standardised pipelines, engineering docs and release processes are the cheapest way I know to make a small team deliver consistently — it’s the first thing I set up on every team I join.