Can Dedeoglu — AI Agents & Complex Solutions

Professional Summary

I architect and build complex AI-powered systems — from multi-agent orchestration pipelines and vision-capable RAG solutions to enterprise-scale automation frameworks that connect disparate platforms and make them work as one. With 18+ years at Microsoft and a hands-on development background across TypeScript, Python, and C#, I close the gap between AI research and production reality: I don't just design, I ship.

My focus is building what others consider impossible: agent-to-agent systems that self-coordinate, secure MCP architectures that comply with enterprise governance, and complex integrations — like a healthcare user-lifecycle automation serving 20,000+ users or a secure B2B file-sharing platform — that require both architectural breadth and developer depth. I lead from the technical front, translate ambiguity into elegant systems, and mentor teams along the way.

AI Agent Orchestration Complex Solutions Design Advanced RAG & Vision Models Enterprise Automation MCP & LLM Security

Core Competencies

AI Agents & Automation

Multi-agent & agent-to-agent orchestration (MAF, Semantic Kernel, LangChain)
Custom Copilot Studio & M365 Agents SDK integrations
Advanced RAG pipelines with vision model support
MCP server design secured with OAuth2 / Zero Trust
Prompt engineering, LLM cost optimization, hallucination reduction
Azure OpenAI, Azure AI Foundry, Open-source LLMs (Llama, Qwen, Kimi, MiniMax, etc.)

Complex Solutions & Architecture

End-to-end complex system design: B2B SaaS, secure file platforms, insurance mgmt
Enterprise IT automation — user lifecycle, identity federation, data pipelines
Microservices, CI/CD, Docker
Azure infrastructure & security architecture (Zero Trust)
Cross-platform integration: Oracle Fusion, Active Directory, Microsoft 365
TypeScript · Python · C# · PowerShell · React / Next.js

Complex Solutions Built

A selection of end-to-end systems designed and developed personally:

Healthcare User Lifecycle Automation — Architected and developed a fully automated identity orchestration system bridging Oracle Fusion HCM and Active Directory for a healthcare enterprise with 20,000+ users. Zero-touch provisioning, role-based access, and real-time sync across systems.

Secure B2B File Sharing Platform — Designed and built an enterprise-grade document exchange system with end-to-end encryption, granular access controls, audit trails, and compliance guardrails for regulated industries.

B2B Insurance Management System — Delivered a multi-party insurance lifecycle platform handling policy issuance, claims routing, and broker integrations across enterprise boundaries.

AI Agent Automation Fleet — Helped 100+ enterprise customers design and develop production AI automation agents covering IT ops, HR workflows, compliance monitoring, and knowledge retrieval — many using vision models and advanced RAG over proprietary data.

Enterprise MCP Security Framework — Designed the pattern for securing Model Context Protocol servers with OAuth2 Proxy in Microsoft's ecosystem, enabling safe AI tool discovery at scale across 40+ enterprise customers.

RAG Hallucination Reduction at Scale — Implemented AI Search + RAG architectures that reduced LLM hallucination rates ~40% for production workloads in critical industries.

Professional Experience

Senior Cloud Solution Architect — AI, Copilot & Complex Solutions Jul 2022 – Present

Microsoft · Dubai, UAE

Drive AI-powered digital transformation across 40+ strategic enterprise customers in the Middle East, owning end-to-end architecture and delivery of complex, custom solutions.
Lead architect and developer for 15+ enterprise Copilot integrations using Copilot Studio and M365 Agents SDK — spanning agent-to-agent systems, advanced RAG, and vision model pipelines.
Designed and built the enterprise MCP server security pattern (OAuth2 Proxy) now applied across the Microsoft Middle East ecosystem for secure AI tool discovery.
Developed production-grade AI agents in TypeScript, Python, and C# using Semantic Kernel, MAF, and Azure AI Foundry for complex multi-step reasoning and automation tasks.
Conducted LLM cost optimization assessments achieving up to 30% token reduction via intelligent caching and model routing strategies.

Technical Account Architect & Premier Field Engineer 2010 – 2022

Microsoft · UAE

Trusted technical advisor to premier enterprise customers; led complex hybrid identity migrations (100,000+ users), regional PowerShell automation frameworks for AD/Exchange, and high-stakes architectural risk assessments across banking, government, and healthcare.

Senior Support Engineer — Exchange & Messaging 2008 – 2010

Microsoft · Istanbul, Türkiye

Mission-critical escalation management for Exchange Server enterprise deployments, resolving complex high-pressure incidents and building deep expertise in Microsoft messaging infrastructure.

Thought Leadership — candede.com

I publish weekly in-depth architectural guides on AI agents, MCP security, enterprise integration, and complex solutions design. Topics covered include:

Enterprise MCP OAuth2 identity patterns

Agent-to-agent orchestration architectures

Advanced RAG with vision model integration

Governance & control plane design for AI agents

LLM vulnerability & prompt injection defense

Microsoft AI SDK comparisons and selection guides

TypeSpec & API-first development for Copilot

Practical LLM cost optimization strategies

Certifications · Education · Languages