AI Architect — the operator who also draws the system
An AI architect decides what the model is allowed to know, what it may do, and how the pieces stay cheap and honest as they grow. I have designed those layers for every platform I run, and I have sat in the boardroom that has to live with the result. Both seats, one person.
What an AI architect decides
Three decisions outweigh all others. The knowledge layer: where truth lives and how the model reaches it — on this site, both the Brain and the Dictionary answer from a single server-side knowledge graph, so there is one place to correct a fact. The agent layer: which tasks get their own agent, how agents hand off, and what a human must approve — the Metals Trading Desk runs several agents across three market sessions with a signal brief as the only output a reader sees. The cost and trust layer: caching, lazy loading, evaluation sets, and the audit trail regulators will ask for.
Architectures I have shipped
- One graph, two products. khurrambadar.com serves a site and a 13,000-term dictionary from the same knowledge graph via serverless functions on Vercel — no duplicated content, one source of truth.
- Autonomous school. newworld.education: curriculum-aligned generation with a verification pipeline — 89,000+ questions verified before students ever see them.
- Multi-agent desk. metalstradingdesk.com: technicals, macro and physical-premium agents feeding one daily brief per session.
- Agent factory. createagent.ai: a platform for creating and deploying AI agents rather than hand-building each one.
- Sixteen-language surface. One message recorded once, translated with my own face and voice by AI video; site and assistant respond in all sixteen, with Arabic and Urdu in full right-to-left.
Why the operator matters
Founder of Spotlight FZE in Dubai Media City in 2002. COO of SkyH2O (atmospheric water, UAE and Singapore). Part of the STRATEMA team behind a $110M solar-hydroponic campus for Masdar, Silal and Alpha Dhabi. ESG advisory under UAE Federal Decree-Law No. 11/2024. An architecture that ignores how a company actually approves spending, hires and reports is a diagram, not a system. I design for the company as it is.
Questions people ask
What is the difference between an AI architect and an AI engineer?
The architect decides the shape: knowledge sources, agent boundaries, data flows, governance and cost. The engineer builds and maintains the pipelines inside that shape. I do both — see the AI Engineer page — but the architecture decision comes first and is the one boards should care about.
Can you review an architecture our vendor proposed?
Yes. The Convergence Audit includes an architecture review: where truth lives, what the model can touch, what happens when it is wrong, and what it will cost at ten times the volume.
Do you work with our existing cloud and data stack?
Yes. My own platforms run on Vercel serverless functions with commercial model APIs, but the architecture principles apply to any stack; the design is model- and cloud-agnostic by intent.
WhatsApp +971 55 623 9111 · khurrambadar@gmail.com · Dubai