https://yapingyang.github.io/Recent content on yapingyangHugoen-usSat, 01 Aug 2026 16:54:12 -0400https://yapingyang.github.io/posts/attention-kv-cache-and-prompt-caching-explained-from-first-principles/Sat, 01 Aug 2026 16:54:12 -0400https://yapingyang.github.io/posts/attention-kv-cache-and-prompt-caching-explained-from-first-principles/<h2 id="what-is-an-attention-mechanism-and-what-does-it-solve">What Is an Attention Mechanism, and What Does It Solve?</h2> <p>A large language model cannot store the word &ldquo;cat&rdquo;, only numbers. So every word has to become a list of numbers, which we&rsquo;ll call a vector, looked up from a big table. But here&rsquo;s the catch, that lookup is context-free, meaning &ldquo;it&rdquo; in the sentence &ldquo;the cat eats because it was hungry&rdquo; has the same vector representation &ldquo;it&rdquo; gets in every other sentence. On its own, it knows nothing about the cat.</p>https://yapingyang.github.io/about/Sun, 26 Jul 2026 00:00:00 +0000https://yapingyang.github.io/about/<h2 id="applied-ai-engineer--full-stack-software-engineer">Applied AI Engineer &amp; Full-Stack Software Engineer</h2> <p>I build AI-powered applications - RAG pipelines, agent workflows, and automation systems - with a focus on making them reliable, maintainable, and ready for production.</p> <p>Previously, I worked on growth engineering and scaling GTM systems, where I learned to value architectures that hold up under real-world pressure—not just in demos. Today I apply that same systems mindset to building AI products for GTM and beyond.</p>