<?xml version="1.0" encoding="UTF-8" ?>
<rss version="2.0">
    <channel>
      <title>WyrdWerk Deployment Wiki</title>
      <link>https://wiki.wyrdwerk.com</link>
      <description>Last 10 notes on WyrdWerk Deployment Wiki</description>
      <generator>Quartz -- quartz.jzhao.xyz</generator>
      <item>
    <title>Embeddings</title>
    <link>https://wiki.wyrdwerk.com/concepts/embeddings</link>
    <guid>https://wiki.wyrdwerk.com/concepts/embeddings</guid>
    <description><![CDATA[ The vector layer that makes semantic search possible — how text becomes numbers, and why the embedding model you pick determines what your RAG system can actually find. ]]></description>
    <pubDate>Thu, 11 Jun 2026 00:00:00 GMT</pubDate>
  </item><item>
    <title>Fine-Tuning</title>
    <link>https://wiki.wyrdwerk.com/concepts/fine-tuning</link>
    <guid>https://wiki.wyrdwerk.com/concepts/fine-tuning</guid>
    <description><![CDATA[ Teaching a model new behaviors by updating its weights — when it's the right call over RAG, and when it isn't. ]]></description>
    <pubDate>Thu, 11 Jun 2026 00:00:00 GMT</pubDate>
  </item><item>
    <title>Model Quantization</title>
    <link>https://wiki.wyrdwerk.com/concepts/model-quantization</link>
    <guid>https://wiki.wyrdwerk.com/concepts/model-quantization</guid>
    <description><![CDATA[ How reducing model weight precision trades quality for speed and memory — the math behind running 70B models on consumer hardware. ]]></description>
    <pubDate>Thu, 11 Jun 2026 00:00:00 GMT</pubDate>
  </item><item>
    <title>Prompt Engineering</title>
    <link>https://wiki.wyrdwerk.com/concepts/prompt-engineering</link>
    <guid>https://wiki.wyrdwerk.com/concepts/prompt-engineering</guid>
    <description><![CDATA[ Designing AI instructions that behave consistently in production — system prompts, few-shot examples, and the maintenance reality nobody talks about. ]]></description>
    <pubDate>Thu, 11 Jun 2026 00:00:00 GMT</pubDate>
  </item><item>
    <title>Hallucination Failure</title>
    <link>https://wiki.wyrdwerk.com/failures/hallucination-failure</link>
    <guid>https://wiki.wyrdwerk.com/failures/hallucination-failure</guid>
    <description><![CDATA[ When the model generates confident, wrong answers from its training weights — distinct from retrieval failure, and harder to catch. ]]></description>
    <pubDate>Thu, 11 Jun 2026 00:00:00 GMT</pubDate>
  </item><item>
    <title>Prompt Injection</title>
    <link>https://wiki.wyrdwerk.com/failures/prompt-injection</link>
    <guid>https://wiki.wyrdwerk.com/failures/prompt-injection</guid>
    <description><![CDATA[ When user-controlled input hijacks your agent's behavior — the security failure that's invisible until the damage is done. ]]></description>
    <pubDate>Thu, 11 Jun 2026 00:00:00 GMT</pubDate>
  </item><item>
    <title>Vendor Lock-In</title>
    <link>https://wiki.wyrdwerk.com/failures/vendor-lock-in</link>
    <guid>https://wiki.wyrdwerk.com/failures/vendor-lock-in</guid>
    <description><![CDATA[ When switching AI providers, platforms, or tools costs more than staying — and you don't realize it until the bill arrives. ]]></description>
    <pubDate>Thu, 11 Jun 2026 00:00:00 GMT</pubDate>
  </item><item>
    <title>Flowise</title>
    <link>https://wiki.wyrdwerk.com/tools/flowise</link>
    <guid>https://wiki.wyrdwerk.com/tools/flowise</guid>
    <description><![CDATA[ Low-code agent builder for LLM workflows — purpose-built for multi-agent orchestration, RAG pipelines, and LangChain-native deployments. ]]></description>
    <pubDate>Thu, 11 Jun 2026 00:00:00 GMT</pubDate>
  </item><item>
    <title>Langfuse</title>
    <link>https://wiki.wyrdwerk.com/tools/langfuse</link>
    <guid>https://wiki.wyrdwerk.com/tools/langfuse</guid>
    <description><![CDATA[ LLM observability for production deployments — traces, cost tracking, and LLM-as-judge evaluations that catch drift before your users do. ]]></description>
    <pubDate>Thu, 11 Jun 2026 00:00:00 GMT</pubDate>
  </item><item>
    <title>LiteLLM</title>
    <link>https://wiki.wyrdwerk.com/tools/litellm</link>
    <guid>https://wiki.wyrdwerk.com/tools/litellm</guid>
    <description><![CDATA[ The unified API gateway for multi-model AI stacks — route, fallback, and track costs across 100+ LLM providers from one endpoint. ]]></description>
    <pubDate>Thu, 11 Jun 2026 00:00:00 GMT</pubDate>
  </item>
    </channel>
  </rss>