I'm Mehmet. I specialize in LLM tooling, adaptive learning systems, and cognitive-inspired AI. Focused on shipping robust, cloud-native products that solve real problems.
Authoring and editing technical content focused on AI architecture and software engineering. I break down complex models into practical, actionable insights for developers.
Designing systems inspired by human learning dynamics—focusing on attention mechanisms, memory retention, and rigorous model evaluation.
AWS-certified cloud architect building scalable, modern AI applications.
FlagshipA prompt optimization engine built to slash LLM token costs without degrading output quality. Features a local heuristic engine, RAG context management, and a seamless Chrome extension.
An adaptive learning management system leveraging cognitive AI. It utilizes metacognitive tracking and dynamic content generation to tailor education to individual learning dynamics.
Builder-first AI workspace for chat, agent authoring, orchestration jobs, saved workflows, and RAG-backed documents. Features a full ADK workbench with agent lifecycle management, multi-agent coordinator, and local ChromaDB-powered retrieval — all wrapped in an Electron desktop shell.
Educational web demo that transforms natural-language scenarios into brain activity simulations. An LLM classifies brain lobes and neuromodulators, a Brian2 spiking neural network runs the simulation, and a Three.js-powered 3D brain model visualizes the neural activity in real time.
Transforms any Python codebase into an interactive call graph with node-level AI explanations, learning path suggestions, and real-time 'ghost runner' execution visualization. Designed for developers who learn by vibing with code but need structure.