I architect LLM systems that ship into regulated banking.
AI Architect and Head of Engineering in Dubai. I own end-to-end RAG and agentic architecture on sovereign cloud infrastructure — model-agnostic, audited, and running in production for core banking.
Request flow through the sovereign AI platform: Client → model-agnostic gateway → vector retrieval → agentic orchestration → knowledge graph → identity and access → Response.
What I do
I design the architecture that lets enterprises run generative AI in environments where mistakes are not an option. That means provider-agnostic LLM gateways, retrieval and agentic patterns that survive production load, and governance — observability, audit trails, human-in-the-loop — treated as first-class, not afterthoughts. My background is low-level: reverse engineering and systems work. I bring that precision to AI platforms.
Selected work
Sovereign RAG for core banking
A production retrieval system unifying structured and unstructured data inside a regulated banking suite.
Read →Model-agnostic LLM gateway
Provider-independent routing so the platform swaps models without touching application code.
Read →Agentic workflows, safely
Durable, auditable agent execution with guardrails against the failure modes that kill enterprise pilots.
Read →