1
Media Buying
Your ads, placed where your customers actually look. I manage campaigns across Google, Meta, and display networks — turning ad spend into measurable revenue, not just impressions.
I turn data into decisions and campaigns into revenue. With 22 years in digital marketing — from managing MYR 150K+ monthly ad budgets for insurance giants to raising MYR 7 million for social causes — I've learned that great marketing lives at the intersection of analytical rigor and creative instinct. I build strategies grounded in real numbers, not hunches.
Years of experience
Your ads, placed where your customers actually look. I manage campaigns across Google, Meta, and display networks — turning ad spend into measurable revenue, not just impressions.
More conversions from the traffic you already have. I analyze your funnel, test what works, and fix what doesn't — because driving traffic to a leaky bucket is just expensive water play.
Qualified leads that actually pick up the phone. Using AI-powered targeting and proven nurture sequences, I build pipelines that fill your CRM with prospects ready to buy — not just browse.
New markets, strategic partnerships, and revenue channels you haven't tapped yet. I help you identify growth opportunities and build the relationships that turn them into real business.
Dashboards that tell you what to do, not just what happened. I turn your raw data into clear, actionable insights — so every marketing decision is backed by evidence, not guesswork.
Custom tools that save hours, not create headaches. From marketing automation scripts to full web applications, I build solutions that make your processes faster and your team more productive.

I run a freelance agency and a SaaS side project from Malaysia. My main PC handles everything when I am at my desk, but I occasionally work from coffee shops, co-working spaces, and on the road. I needed a dedicated Linux machine for mobile work — something that could handle real development workloads without being tethered to a charger. This is the story of how I searched for, evaluated, bought, and upgraded a refurbished ThinkPad T480, and then solved the harder problem: making my entire AI-assisted development environment portable between two machines with a single git pull.

Earlier this year, the economics of AI compute changed. Major providers — Anthropic, OpenAI, Google — started adjusting pricing upward after years of subsidising adoption. The demand-side story is simple: inference capacity is not keeping pace with how fast developers are integrating these tools into their daily workflows. For anyone running a serious amount of AI-assisted work, the cost trajectory is no longer abstract. I run a web development agency, and I was coding against multiple AI providers daily: Claude Code for terminal and config work, GitHub Copilot inside the IDE, Codex CLI for secondary coding runs. When I sat down and calculated what that looked like at scale — input tokens from behavioral rules loaded on every session, MCP tool schemas injected into every context window, 8 servers active simultaneously when I only needed 1 — the waste was significant and entirely fixable. This post documents exactly what I built to address it: a model-agnostic AI setup where every provider operates from the same behavioral rules, token usage is deliberately managed, and switching models costs me thirty minutes rather than a full rewrite.