--- title: "Muratcan Koylan - Air Market" description: "Muratcan Koylan publishes 16 listings on Air Market. I enjoy building stuff with AI" canonical: "https://airmarket.ai/@muratcankoylan" --- [Air Market](/) / [Creators](/creators) / Muratcan Koylan Creator # Muratcan Koylan I enjoy building stuff with AI **Muratcan Koylan on Air Market** | | | | --- | --- | | Listings | 16 | | GitHub | [muratcankoylan](https://github.com/muratcankoylan) | **Published listings, with what each one does** | Listing | Kind | What it does | | --- | --- | --- | | [Latent Briefing and KV Cache Memory Sharing](/app/latent-briefing) | Skill | In a setup where one agent hands work to others, the same context gets passed around again and again and the bill climbs. This is about pas… | | [Reasoning Trace Optimizer](/app/reasoning-trace-optimizer) | Skill | Sometimes an agent fails and the logs do not tell you why. This looks at its own reasoning to find the cause: it lost track of the instruct… | | [Multi-Agent Architecture Patterns](/app/multi-agent-patterns) | Skill | Splitting work across several agents sounds obvious and often makes things worse. This helps you decide whether you actually need more than… | | [Evaluation Methods for Agent Systems](/app/evaluation) | Skill | When you change an agent, it is hard to tell whether you improved it or broke something else. This is about building the tests that tell yo… | | [Digital Brain](/app/digital-brain) | Skill | Your notes, your contacts and your commitments usually live in different places, so nothing is ready when you need it. This handles the per… | | [Context Optimization Techniques](/app/context-optimization) | Skill | An assistant that reads too much costs more and starts to wander. This helps you decide what it keeps in front of it and what it drops, so… | | [Context Degradation Patterns](/app/context-degradation) | Skill | Sometimes an agent works well and then quietly stops working well, without anything obvious changing. Often the reason is what has piled up… | | [Book SFT Pipeline](/app/book-sft-pipeline) | Skill | If you want a model to write in a particular author's voice, you need that author's writing turned into training material first. This walks… | | [BDI Mental State Modeling](/app/bdi-mental-states) | Skill | There is an old and careful way of describing a thinking agent: what it believes, what it wants, and what it has committed to doing. This h… | | [Tool Design for Agents](/app/tool-design) | Skill | An agent picks the wrong tool mostly because the tools were described badly. This is about that layer: what each tool is called, what it ta… | | [Project Development Methodology](/app/project-development) | Skill | Early choices on a language-model project are the expensive ones to undo. This is for those: whether a model is even the right piece here,… | | [Memory System Design](/app/memory-systems) | Skill | Most agents start every session blank, so you explain the same things over and over. This is about giving one a memory that lasts: what it… | | [Long-Horizon Prompting](/app/long-horizon-prompting) | Skill | Setting an agent loose on a hard problem for hours goes wrong mostly in the first paragraph you gave it. This helps you write that brief: w… | | [Context Engineering Fundamentals](/app/context-fundamentals) | Skill | Everything people do to improve an agent comes back to what sits in its context window, and it is hard to make good calls there without und… | | [Context Compression Strategies](/app/context-compression) | Skill | When a working session runs long, the important parts get buried in everything else. This condenses it into a summary that keeps the decisi… | | [Advanced Evaluation](/app/advanced-evaluation) | Skill | When a language model writes something, deciding whether it is good is the hard part. This helps you set up automatic grading: scoring answ… | ## Links - [muratcankoylan.com](https://www.muratcankoylan.com/) - [GitHub](https://github.com/muratcankoylan) --- Canonical URL: