<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Aki&apos;s Tech Garden</title><description>A living collection of experiments, useful systems, and honest notes from the space between research and production.</description><link>https://akilesh.pages.dev/</link><language>en</language><item><title>Architechural Masterclass in Kimi K3</title><link>https://akilesh.pages.dev/writing/architechural-masterclass-in-kimi-k3/</link><guid isPermaLink="true">https://akilesh.pages.dev/writing/architechural-masterclass-in-kimi-k3/</guid><description>A 2.8T parameter model is impressive!!
But what interested me more was, how do you even run this thing?
This is my small attempt at unpacking the architecture choices that make Kimi K3 actually survive in production.</description><pubDate>Thu, 20 Aug 2026 00:00:00 GMT</pubDate><category>Field Notes</category></item><item><title>What a benchmark cannot tell you</title><link>https://akilesh.pages.dev/writing/what-a-benchmark-cannot-tell-you/</link><guid isPermaLink="true">https://akilesh.pages.dev/writing/what-a-benchmark-cannot-tell-you/</guid><description>On the gap between a passing evaluation and a useful product.</description><pubDate>Thu, 06 Aug 2026 00:00:00 GMT</pubDate><category>Research</category><category>Field Notes</category></item><item><title>Building the smallest version that can teach you something</title><link>https://akilesh.pages.dev/writing/smallest-version-that-teaches/</link><guid isPermaLink="true">https://akilesh.pages.dev/writing/smallest-version-that-teaches/</guid><description>A practical way to separate an experiment from a disguised production system.</description><pubDate>Tue, 28 Jul 2026 00:00:00 GMT</pubDate><category>Builds</category><category>Experiments</category><category>Field Notes</category></item><item><title>Cache Poisoning</title><link>https://akilesh.pages.dev/writing/cache-poisoning/</link><guid isPermaLink="true">https://akilesh.pages.dev/writing/cache-poisoning/</guid><description>Yo!! A quick note before we start. I have yapped a lot about prompt caching in Part 1 , how it works, what cache hits and misses actually mean, and how to improve cache hit rates without making your API bill shoot the sky.</description><pubDate>Tue, 23 Jun 2026 00:00:00 GMT</pubDate><category>Research</category><category>Builds</category></item><item><title>Prompt Caching</title><link>https://akilesh.pages.dev/writing/prompt-caching/</link><guid isPermaLink="true">https://akilesh.pages.dev/writing/prompt-caching/</guid><description>Okay so real talk before we start….</description><pubDate>Tue, 09 Jun 2026 00:00:00 GMT</pubDate><category>Research</category><category>Future Thinking</category></item><item><title>AI Does Multiplication Underneath. So Why Did Older Models Break at School Maths?</title><link>https://akilesh.pages.dev/writing/ai-does-multiplication-underneath-so-why-did-older-models-break-at-school-maths/</link><guid isPermaLink="true">https://akilesh.pages.dev/writing/ai-does-multiplication-underneath-so-why-did-older-models-break-at-school-maths/</guid><description>Like anyone who got pulled into the AI wave, I went through the whole thing. The amazement phase, the over reliance phase, the okay this thing just hallucinated phase.., and then the slow ohh okay I think I am starting to get what this actu</description><pubDate>Tue, 19 May 2026 00:00:00 GMT</pubDate><category>Research</category><category>Builds</category></item><item><title>The Smallest Thing in PyTorch Opens Half the GPU Stack</title><link>https://akilesh.pages.dev/writing/the-smallest-thing-in-pytorch-opens-half-the-gpu-stack/</link><guid isPermaLink="true">https://akilesh.pages.dev/writing/the-smallest-thing-in-pytorch-opens-half-the-gpu-stack/</guid><description>We are living in a time when AI systems are introduced with the kind of language people used to reserve for moon missions.</description><pubDate>Mon, 30 Mar 2026 00:00:00 GMT</pubDate><category>Research</category><category>Builds</category></item><item><title>What Happens When You Put “n” Billion Weights in Your RAM</title><link>https://akilesh.pages.dev/writing/what-happens-when-you-put-n-billion-weights-in-your-ram/</link><guid isPermaLink="true">https://akilesh.pages.dev/writing/what-happens-when-you-put-n-billion-weights-in-your-ram/</guid><description>I was in full vibe-coding mode with Headphones on. Letting Copilot autocomplete half my thoughts. Prompt here, tab there. Confidence at an all-time high. It honestly felt like I had rented extra IQ from the cloud ( having this kind of feel</description><pubDate>Fri, 27 Feb 2026 00:00:00 GMT</pubDate><category>Research</category><category>Builds</category></item><item><title>Reading Minds Is the New Logging</title><link>https://akilesh.pages.dev/writing/reading-minds-is-the-new-logging/</link><guid isPermaLink="true">https://akilesh.pages.dev/writing/reading-minds-is-the-new-logging/</guid><description>We are in that phase of software engineering where we churn out code at an alarming speed. You open an AI-powered IDE, type a sentence ( technically a prompt ) that sounds more like a thought than a specification, and suddenly, there are hu</description><pubDate>Thu, 22 Jan 2026 00:00:00 GMT</pubDate><category>Research</category><category>Future Thinking</category></item><item><title>The Physics of Prompting: Why Words Don’t Matter, But Structure Does</title><link>https://akilesh.pages.dev/writing/the-physics-of-prompting-why-words-dont-matter-but-structure-does/</link><guid isPermaLink="true">https://akilesh.pages.dev/writing/the-physics-of-prompting-why-words-dont-matter-but-structure-does/</guid><description>For the past few months, I’ve basically been living inside this sacred AI vibing + building multi agent systems + prompting zone. You know that phase where your YouTube is full of ‘how to prompt better’ videos, your tabs are full of blog po</description><pubDate>Sat, 06 Dec 2025 00:00:00 GMT</pubDate><category>Research</category><category>Future Thinking</category></item><item><title>Catastrophic Forgetting: The Architecture of Becoming Human</title><link>https://akilesh.pages.dev/writing/catastrophic-forgetting-the-architecture-of-becoming-human/</link><guid isPermaLink="true">https://akilesh.pages.dev/writing/catastrophic-forgetting-the-architecture-of-becoming-human/</guid><description>I lost something this week and I still don’t know how to say it. Not my car, not a memory. Worse than that. I lost how I think….</description><pubDate>Tue, 25 Nov 2025 00:00:00 GMT</pubDate><category>Research</category><category>Future Thinking</category></item><item><title>Latency Diet for LLMs: Cutting Millisecond Fat Without Losing IQ</title><link>https://akilesh.pages.dev/writing/latency-diet-for-llms-cutting-millisecond-fat-without-losing-iq/</link><guid isPermaLink="true">https://akilesh.pages.dev/writing/latency-diet-for-llms-cutting-millisecond-fat-without-losing-iq/</guid><description>When I first started building my own model ( a language model for my experimental need ), I went through a lot of videos and built it from scratch. While doing so I noticed something interesting: Speed feels like intelligence. A model that</description><pubDate>Mon, 13 Oct 2025 00:00:00 GMT</pubDate><category>Research</category><category>Builds</category></item></channel></rss>