A space where my thoughts take flight.https://pinglin.tw/en-ushttps://pinglin.tw/blog/collection-autofill-at-scale/https://pinglin.tw/blog/collection-autofill-at-scale/A design walkthrough of how Collection Autofill — a spreadsheet of AI-computed cells — stays fast, fair, and observable from a hundred rows to ten million.Thu, 28 May 2026 00:00:00 GMThttps://pinglin.tw/blog/reimagine-unstructured-data-etl/https://pinglin.tw/blog/reimagine-unstructured-data-etl/Generative AI is revolutionizing software—how does it transform the way we utilize unstructured data like text, images, and videos?Fri, 14 Feb 2025 00:00:00 GMThttps://pinglin.tw/blog/the-shapes-of-agent-memory/https://pinglin.tw/blog/the-shapes-of-agent-memory/An agent that remembers across sessions can keep its memory as curated markdown files, as an auto-mined structured store, or as trained experience. I measured all of them: files against a structured store under one fixed model, a store-only head-to-head across the structured lineages, and an experience bank on the agentic benchmarks where the state of the art trains memory into the weights.Wed, 12 Aug 2026 00:00:00 GMThttps://pinglin.tw/blog/why-steering-works/https://pinglin.tw/blog/why-steering-works/An agent run is stochastic gradient descent over solution space, and steering is how the true objective enters the loop. An optimization view and a Bayesian view that ground the industry ladder — prompt, context, harness, loop engineering — in foundation theory, and explain when a big model steering a small one beats distillation, and when it does not.Sun, 26 Jul 2026 00:00:00 GMThttps://pinglin.tw/blog/optimizing-apple-silicon-gpu-for-transformer-inference/https://pinglin.tw/blog/optimizing-apple-silicon-gpu-for-transformer-inference/One MacBook Pro M4 Max, an open-source MoE, and a hands-on exercise in optimizing it for inference: why the thing that collapses LLM throughput by 100× is prefill, not decode, and the two prefill-side fixes that got it back.Sat, 11 Jul 2026 00:00:00 GMT