The garden index
Writing
Notes from research, building, and the useful friction between a clean idea and a working system.
All writing
Architechural Masterclass in Kimi K3
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.
What a benchmark cannot tell you
On the gap between a passing evaluation and a useful product.
Building the smallest version that can teach you something
A practical way to separate an experiment from a disguised production system.
Cache Poisoning
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.
AI Does Multiplication Underneath. So Why Did Older Models Break at School Maths?
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
The Smallest Thing in PyTorch Opens Half the GPU Stack
We are living in a time when AI systems are introduced with the kind of language people used to reserve for moon missions.
What Happens When You Put “n” Billion Weights in Your RAM
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
Reading Minds Is the New Logging
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
The Physics of Prompting: Why Words Don’t Matter, But Structure Does
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
Catastrophic Forgetting: The Architecture of Becoming Human
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….
Latency Diet for LLMs: Cutting Millisecond Fat Without Losing IQ
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