Architecting an LLM-native platform for Naksha, a greenfield desktop app
that compiles a plain-language brief into architectural drawings through an
LLM agent loop and a deterministic Python engine.
-
Design and maintain the LLM serving stack end-to-end using Cloudflare
Workers and AI Gateway following a hardened hub and spoke model
-
Build infrastructure and automation for cost controls and per-seat
licensing
-
Write a custom provider-agnostic serving layer against the OpenAI API spec
for seamless model changes
-
Build an automated evals harness for traceable prompt and model changes
-
Build CI/CD pipelines in GitHub Actions for our Tauri sidecar system, with
canary releases and auto-publishing artefacts to Cloudflare R2
Technical architect and developer on miXai-learn, a ¥20.9M KAKENHI-funded
international project developing socially assistive robotics for embodied
learning.
-
Design and construct on-premises server infrastructure for the LA-ReflecT
research platform, including Docker Swarm orchestration, Cloudflare
integration, and nginx, serving distributed research teams across three
Japanese universities
-
Build a live-in ChatOps agent on open-weight models with Ollama and
LangGraph, embedded in Mattermost for remote server debugging, with
guardrailed tool use and audit logging
-
Lead the design and delivery of the robot integration system, enabling
socially assistive robots to support interactive classroom learning
through LLM-driven gesture selection and real-time hardware telemetry,
with session photography and conversation transcripts feeding the
LA-ReflecT analytics platform
-
Mentor academic researchers in DevOps culture and enterprise practices,
leading codebase migration from GitLab to GitHub with CI/CD pipelines,
and establishing contribution protocols and version control standards
across partner universities in Japan and Europe
Holding the reins of DevOps & Infrastructure in production,
while providing a nourishing and dependable atmosphere for
all technical and non-technical teams.
-
Design and construct the future infrastructure of Aotal,
in collaboration with internal stakeholders
-
Manage the complete lifecycle of Aotal's Helm - Kubernetes
- Terraform - AWS backend single-handedly, resolving technical debt accumulated over years
-
Architect and execute comprehensive infrastructure migration
from legacy EKS to AWS Fargate across 50+ services, reducing
infrastructure costs by ~33%, eliminating recurring quarterly
upgrade disruption, and enabling developer self-service through
a full Terraform rewrite
-
Work fully remotely with the development and operations
teams in New Zealand
-
Provide 24/7 on-call support for Aotal's critical systems
serving clients across New Zealand, Australia and the USA
-
Mentor the Auckland based operations team with DevOps best
practices and guide them through legacy migrations
Leading operational DevOps and rapport with four contrasting
teams in Stuff's Auckland office solo.
-
Ensure the highly available stuff.co.nz website serving
over two million customers every day remains available and
pristine
-
Work remotely with the wider DevOps team, while working
closely with developers in the Auckland office
-
Rebuild the DevOps team's reputation across four product
teams with distinct tech stacks, establishing it as a
reliable and trusted engineering partner
-
Mentor and support juniors in technical best practices,
and operational protocols
-
Build and maintain Stuff's critical legacy infrastructure
in AWS using Docker, Puppet and Concourse
-
Build and maintain Stuff's new infrastructure in GCP using
Kubernetes, Jenkins and Terraform
-
Configure and maintain a highly sophisticated CDN network
using Akamai and Varnish caching
-
Provide on-call support utilizing a NewRelic and ELK based
monitoring stack over PagerDuty
-
Design and deliver a custom infrastructure pipeline for
the New Product Development team, built around their
divergence from the existing codebase
Grew from graduate engineer to DevOps culture driver across
Gentrack's global product teams, spanning enterprise software
delivery and automation infrastructure.
-
Build a CI pipeline from scratch using Python, AWS and
Jenkins, establishing the primary benchmarking tool for
all production releases
-
Create a test harness extension to the CI pipeline for
customizing automated performance tests written for JMeter
-
Develop enterprise-level Billing and CRM software in Java
Spring Hibernate and C#, including an SFTP integration
with Windows Component Services, delivered for the
Australian and Singapore electricity market
-
Break down locally-hosted monolithic programs into
cloud-hosted micro-services with deployment automation
using Terraform and best security practices
-
Build a SaaS application to provision CI pipelines across
multiple products using Jenkins Job DSL, AWS Boto3 and
Terraform
-
Guide product teams in adopting DevOps as a culture;
mentor and lead interns in producing a proof-of-concept
SaaS application