About
I'm a software engineer in Auckland. I've been building backends for about ten years, mostly for systems that have to answer while someone is waiting.
Right now I'm at Simbie AI, building the Python backend behind AI voice agents that pick up the phone at medical clinics. I also work on TimeTackle, a calendar analytics platform I've been building since 2020.
Before New Zealand I spent seven years at Brain Station 23 in Dhaka and left as a Staff Software Engineer. Along the way I led AXAR AI, an open-source agent framework, and taught React and AWS to two bootcamp cohorts.
I'm finishing an MSc at Unitec on detecting Parkinson's from a voice recording. Same signal I work with all day, different question.
Open to senior backend, ML or AI engineering roles in New Zealand.
Work Experience
Skills
Things I've shipped
A few projects I've worked on recently — an open-source agent framework, production voice agents in healthcare, and integration work on a calendar analytics platform.
Open-source TypeScript framework for building production agents — type-first, model-agnostic, with explicit control instead of hidden magic. 163 stars. I built the release pipeline that publishes to both npm and JSR, added Prettier and coverage to CI, and drove the rename from axar-core to axar across the packaging workflows.
An AI receptionist for medical clinics: answering calls, scheduling appointments and handling patient intake. I work on the Python backend — telephony, EHR integration, and the task queues behind it. Isolating heavy Celery tasks and right-sizing the API workers took call exports from 25-30 minutes down to about one minute.
Calendar analytics and time tracking for teams, on a Java and Spring Boot backend. I have worked on this product since 2020, first through Brain Station 23 and directly since March 2025, and built most of what runs today: the core queue system, reporting and charting, workflow automation, Meeting Health, the tagging engine, and the integrations into Salesforce, HubSpot, Deel, ClickUp and Clio. Along the way I halved the GCP bill by moving services to Cloud Run and roughly doubled Node.js throughput with caching and query work.
An AI sales workspace for people who live in back-to-back calls. I led the build. It ships as a desktop app across every OS, hooks into Zoom, and turns each call into something useful: transcription, then text analysis over what was actually said, then GPT-3.5 Turbo predicting the follow-up and offering it as one click. The point was to cut the admin tail off every meeting.
A US social banking app that put checking, a line of credit and savings behind one balance and paid cash rewards on spending. At Brain Station 23 I built the money transfer module and the rewards program, then added the MeedTravel and MeedCover products on top.
Get in Touch
I'm open to senior backend, platform and AI engineering roles in New Zealand. The fastest way to reach me is email — I read everything and reply to anything that isn't a mass mailout.






