Systems · AI · Platform Engineering

I build systems that turn complexity into momentum.

I work across production AI, platform architecture, and delivery systems—helping ambitious products scale without adding unnecessary complexity.

Say hello See selected work

Production AI
Multi-region platforms
Thoughtful delivery systems

Selected impact

Useful systems make good work easier.

1M+

Daily requests across multi-region platforms

99.99%

Availability at marketplace scale

35%

Infrastructure cost reduction

Domain-aligned services in the TMS transformation

How I work

Make the system clearer for the people inside it.

I create the conditions for teams to make better decisions: clear product context, measurable quality bars, ownership boundaries, and delivery practices that turn strategy into reliable execution.

My job is not to be the smartest person in every room. It is to build the operating system that helps engineers, product partners, and emerging leaders make good calls without waiting for permission.

Direction with context

Translate company goals into a technical strategy, sequence difficult decisions, and make trade-offs visible before they become surprises.

Quality that scales

Build reliability, observability, delivery controls, and ownership into the platform instead of relying on heroics or hidden knowledge.

AI with accountability

Adopt agents where they create real leverage, with evaluation, rollout controls, human handoff, and a business case worth measuring.

Selected leadership work

The systems behind the headline.

Applied AI · Singapore + Australia

Production multi-agent platform

Led a nine-agent platform for customer support, sales, collections, verification, and onboarding. The work was not “add a chatbot”; it was designing an operating model for agents that could be rolled out, evaluated, escalated, and trusted in production.

Role: Engineering strategy, agent-platform architecture, delivery governance
System: 9 specialist agents across two markets with RAG, human handoff, and evaluation
Leadership question: How do we use AI to create leverage without lowering the quality bar?

Claude Agent SDK
MCP
RAG
LLM evals

Platform modernization

Marketplace monolith to TMS

Led the architecture and staged evolution toward a multi-tenant, domain-aligned platform: TypeScript services, gRPC, Kafka with transactional outbox, PostgreSQL, contract testing, and deployment standards.

Role: Platform architecture and engineering standards
Decision: Extract bounded domains while making contracts, events, and delivery quality explicit.

Microservices
Kafka
gRPC
Terraform

Reliability engineering

Safer payments across five systems

Resolved a production race condition across a mini-app, provider webhook, payment ledger, and booking platform by introducing deferred capture, idempotency, post-commit reconciliation, and safer recovery paths.

Role: Incident leadership and reliability design
Decision: Never capture money until both the payment and booking signals have converged.

Payments
Idempotency
Eventual consistency

Developer experience

AI-native delivery workflow

Built an autonomous ticket-to-PR workflow: a Jira event routes through a Lambda and GitHub Actions to an AI coding agent that prepares reviewable implementation work across repositories and regions.

Role: AI-enabled developer experience
Principle: Use AI to shorten feedback loops while keeping review and ownership human.

GitHub Actions
AWS Lambda
OpenAI Codex

Selected experience

A builder's path through systems, products, and platforms.

2024 - Present

Head of Engineering · Drive lah
Engineering strategy, platform evolution, delivery operating model, and AI adoption for a multi-region marketplace.
2020 - 2024

Technical Lead · Drive lah
Multi-region architecture, platform modernization, reliability, and engineering standards.
2016 - 2020

Senior engineering roles · Josudo, Attentive AI, RenewBuy
Backend platforms, AI and GIS services, insurance integrations, automation, and developer tooling.

Notes from the work

The principles I return to when the stakes are high.

These are the ideas I return to when the roadmap is unclear, the system is under pressure, or the next AI initiative needs a real operating model.

Follow my thinking on LinkedIn

01 / Strategy

Start with the constraint, then choose the system.

Good strategy names the problem, the constraint, the decision, and the trade-off. A roadmap is useful only when it changes the team's next decision.

02 / Quality

Reliability is an operating model, not an on-call personality trait.

Ownership, observability, safe delivery, and clear recovery paths are how quality becomes repeatable instead of depending on the same people saving the day.

03 / AI adoption

AI earns its place when it improves a workflow people already care about.

Choose the workflow, define the evaluation, stage the rollout, keep meaningful human control, and measure the outcome. That is how experiments become capability.

Building in public

Small tools. Sharp ideas.

Open source · Python

penprint

A deterministic voice gate for writing: turn “does this sound like me?” into a repeatable score without an LLM, so writing-agent loops can converge.

pip install penprint