<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Suat Erken — Field notes</title><description>Field notes on coding agents, model evaluation, product development, and web performance.</description><link>https://suaterken.dev/</link><language>en</language><item><title>Making Unity legible to coding agents</title><link>https://suaterken.dev/lab/agent-friendly-unity/</link><guid isPermaLink="true">https://suaterken.dev/lab/agent-friendly-unity/</guid><description>How I give coding agents enough access to inspect a Unity change, see the rendered result, and repair obvious mistakes before review.</description><pubDate>Sat, 18 Jul 2026 00:00:00 GMT</pubDate><category>Unity</category><category>MCP</category><category>Agent workflows</category><category>Visual verification</category></item><item><title>A risk-tiered approach to AI adoption</title><link>https://suaterken.dev/lab/ai-adoption-guardrails/</link><guid isPermaLink="true">https://suaterken.dev/lab/ai-adoption-guardrails/</guid><description>A practical way to vary specifications, review, and access according to the cost of getting an AI-assisted change wrong.</description><pubDate>Sat, 18 Jul 2026 00:00:00 GMT</pubDate><category>AI adoption</category><category>Developer experience</category><category>Guardrails</category><category>Engineering leadership</category></item><item><title>A reviewable multimodal production pipeline</title><link>https://suaterken.dev/lab/multimodal-production/</link><guid isPermaLink="true">https://suaterken.dev/lab/multimodal-production/</guid><description>How I organise generated 2D, 3D, audio, voice, and video as a traceable production process rather than a collection of one-off prompts.</description><pubDate>Sat, 18 Jul 2026 00:00:00 GMT</pubDate><category>ComfyUI</category><category>Multimodal AI</category><category>Asset pipelines</category><category>Human review</category></item><item><title>Evaluating open-weight models after fine-tuning</title><link>https://suaterken.dev/lab/open-weight-evaluation/</link><guid isPermaLink="true">https://suaterken.dev/lab/open-weight-evaluation/</guid><description>How I compare LoRA adapters for Qwen, Llama, and Mistral against a held-out task set, an explicit baseline, and the realities of serving them.</description><pubDate>Sat, 18 Jul 2026 00:00:00 GMT</pubDate><category>LoRA</category><category>Open-weight models</category><category>Evaluation</category><category>Inference</category></item><item><title>Building web performance into everyday delivery</title><link>https://suaterken.dev/lab/web-performance/</link><guid isPermaLink="true">https://suaterken.dev/lab/web-performance/</guid><description>How I keep rendering, caching, third-party code, SEO, and Core Web Vitals inside the normal delivery loop on international commerce platforms.</description><pubDate>Sat, 18 Jul 2026 00:00:00 GMT</pubDate><category>Core Web Vitals</category><category>SEO</category><category>Edge caching</category><category>International commerce</category></item></channel></rss>