A. Semsettin Ozdemirden

Delivering Agentic AI systems across reinforcement learning, computer vision, and LLM engineering. Figma RL Environment: verifiable RL environment for teaching LLMs the Figma Plugin API, targeting licensing to OpenAI, Anthropic, and Google. 7-layer reward stack (structural, SSIM, VLM judge, human preference). Frontier models score 20-30%.

EU IVDR compliance platform for diagnostic manufacturers including Roche, Illumina, and Thermo Fisher. Full-stack Angular + Spring Boot in safety-critical regulated environment. Kubernetes with GitOps (Helm, Flux), multi-repo release automation, E2E testing (Nightwatch, Cucumber BDD).

Delivering Agentic AI systems across reinforcement learning, computer vision, and LLM engineering. Figma RL Environment: verifiable RL environment for teaching LLMs the Figma Plugin API, targeting licensing to OpenAI, Anthropic, and Google. 7-layer reward stack (structural, SSIM, VLM judge, human preference). Under Armour: AI pose synthesis pipeline; Stable Diffusion + ControlNet, MediaPipe pose estimation, InsightFace face swap. HerGunMac: AI football analytics platform with 30K+ monthly users; multi-agent pipelines and evaluation. Figma AI Designer: SaaS plugin converting natural language to real editable Figma nodes. Agentic pipeline with Claude Sonnet 4, BAML outputs, and self-verification via screenshots. Multi-Agent Framework: 24/7 production system with browser automation, cron workers, ElevenLabs voice, memory, and shared state. NestNeo (Co-Founder): AI parenting platform with personalized baby medical profiles.