I build AI systems for the real world.
I help technical founders and product leaders turn promising AI ideas into reliable products. I also build enterprise systems where security, cost, and human judgment matter.
the starting point
AI Product Reality Sprint
10 business daysMake the risky part real before you scale it.
We identify the critical workflow, shape the architecture, build a working vertical slice, and expose the cost, security, and reliability risks before you spend months scaling the wrong system.
the one that proves it
enterprise proof · anonymized
A document-understanding platform for a regulated Malaysian bank
Designed, built, and shipped an AI document-understanding platform inside a regulated, data-resident, human-in-the-loop environment, where the model is the easy part and everything around it (residency, auditability, a human who can always say no) is the hard part.
field notes
Notes from building AI systems that have to survive real constraints: cost, governance, reliability, and human judgment.
and I ship, repeatedly
AI automation platform for businesses: automates workflows, scales operations, and drives growth through intelligent agents.
Write a poem: it's painted as a text-free ink-wash scene, sealed, even sung.
Turn long video into clear, structured answers on demand: ask a question, get the exact moment and a summary.
StoryFlow breathes new life into your footage, reworking the narration so your story lands the way you meant it.
Turn a product photo and a brief into a finished campaign: a council of AI agents debates the angle, generates a keyframe and a 9:16 video, then hands back a distribution plan and an ROI hypothesis.
how I work
End-to-end ownership
Whiteboard to production to handover. The measure is whether the client can run it after I leave.
Cost is a design constraint
I track what every system costs to run, and optimize it. I do it for my own portfolio too.
Human-in-the-loop AI
In regulated settings the human is the control, not the inefficiency. Confidence-gated, auditable.