Wake up, visitor._

AI is part of how I build products and lead teams. I build the engineering harness around it: the context, tools, verification, and human review that make AI-assisted work dependable.

At ThoughtFull, I moved engineering to AI-assisted development and built the engineering harness supporting that workflow. I rolled out a company-wide retrieval-augmented knowledge base (RAG), with collaboration and version control through GitHub.

A shorter path from idea to implementation

Before building, we define the goal and the metrics that will tell us whether it worked. Then we build an early MVP, get feedback, and make approval decisions sooner, with AI-assisted development supporting the path from idea to implementation.

I cut unnecessary Scrum ceremonies, keeping only the daily standup. The process became lighter, while clear goals, success metrics, and reviews kept the work accountable.

RAG starts with knowledge worth retrieving

Project history, decisions, research, and past sessions become linked wiki pages with source citations. Engineers and their agents retrieve that knowledge through index and text search, so the next session can build on previous decisions.

Guardrails belong in the architecture

In healthcare, a useful answer must respect the sensitivity of the data behind it. Our internal analytics assistant lets clinical and client teams ask questions in plain English, but selects only from analyst-approved database queries.

The tradeoff is deliberate: less freedom to explore, tighter control over access to mental-health data. Clinical AI remains outside production until named human accountability and clinical sign-off are in place.

The harness makes the workflow dependable

Faster generation is only part of the work. I built the harness around AI-assisted development so shared context, supporting tools, verification, and review are part of the workflow. The goal is to move quickly from idea to MVP and carry that work into production with confidence in what we ship.

Measure before trusting

I compare models for each use case, evaluate their outputs, and test the behavior of model-assisted work. In my own products, I also build agent workflows and MCP servers that connect AI tools to real product capabilities.

AI adoption meant changing how the company works: getting ideas into review earlier, making knowledge easier to retrieve, and keeping people accountable for what ships.

SELECT COUNT(*) FROM demo_bookings
WHERE month = :month;

The customer is an enterprise. The person trusting the product is an individual.

I lead engineering for a digital mental-health platform, with around 15 people across six countries. My remit covers architecture, delivery, security, budget, and the technical side of enterprise client integrations.

Making responsibility concrete

I led ISO 27001 certification and serve as Data Protection Officer. We reduced the AWS bill by 48%. I also built the monitoring and alerting layer, with a named responder for every alert.

I represented ThoughtFull at Datadog Live Malaysia, speaking on AI observability, security, and the changing engineering workflow.

I replaced Carsome's legacy lead-management system and built the team responsible for it.

I designed and shipped the new platform with deep CRM integration. I built the Onboarding team from the beginning, then led it alongside a three-person R&D team as AVP of Engineering.

From design to production

Distributed services, a move from manual deployment to CI/CD, and an incident war-room process. Alongside that, my initiatives included the CARInspect rebuild and AI damage detection with human review.

Several engineers I mentored grew into Engineering Lead roles. The work included developing the people who would lead the next projects.

Retail operations across APAC, built from scratch and rolled out in three months.

As CTO, I oversaw delivery of Lenovo's retail channel management platform: a web management portal and four mobile apps covering promoters, dealers, merchandising, and demo-unit activity.

A platform and a department

I oversaw more than 15 apps and SaaS platforms, rebuilt the technical department of around 20 people, and led the full product portfolio's migration to Google Cloud, designing the target architecture.

I introduced shared engineering standards, trained team leads, and agreed delivery processes with the PMO. The platform rollout and the department rebuild were both part of my remit.