About Ercin

• Sovereign AI platforms. Data residency, identity and audit built into the design from day one. Today this means LLM, RAG and agentic systems that keep bank data inside the country.
• Model-agnostic LLM gateways. Provider-independent routing, so a platform can swap between frontier and open-source models without rewriting application code. See the gateway case study.
• Retrieval and agentic patterns that survive production. RAG over structured and unstructured data, and durable, auditable agent workflows with human-in-the-loop controls. Sovereign RAG for core banking · auditable agentic workflows.
• Defensive AI evaluation. Testing LLM systems against finance-specific risks before go-live and after every model update.

• Governance is architecture. Observability, audit trails and human oversight are first-class components, not afterthoughts.
• Evidence over demos. A system is ready when it holds up under production load and adversarial testing, not when the demo looks clean.
• Low-level precision. My background is systems work and reverse engineering. I bring the same precision to AI platforms.

18+ years of software architecture across banking, aviation, security and embedded systems. Based in Dubai, United Arab Emirates. Open to architecture conversations, advisory and speaking.

Education: Anadolu University.

• Architecting a Sovereign GenAI Platform for Regulated Banking in the UAE: A Reference Architecture — Hugging Face, Jun 2026
• DIFC Regulation 10: The AI Architecture It Demands — Hugging Face, Jun 2026
• All writing →

Case studies: sovereign RAG for core banking, a model-agnostic LLM gateway, durable agentic workflows and security by default →

LinkedIn · Hugging Face · GitHub · X @aiarchitect · Medium · ORCID

Views are my own. Nothing on this site describes confidential systems, clients or employer information.