Ramchand KumaresanR. Kumaresan
ResumeBooksBlogAbout

Blog

Thoughts on AI research, program management, and building products.

GPU SystemsQuantization

The Bug Was Never in the Weights: Quantizing GLM-5.3-Flash to 4 Bits for Two DGX Sparks

A 320B model that does not fit on my hardware, a week spent squeezing it to 4 bits and measuring what that costs on every one of its 37,152 expert tensors, and a corruption bug that turned out to live in one line of vLLM rather than anywhere in the weights.

September 1, 2026
GPU SystemsApple Silicon

The fp8 Mirage: Reverse-Engineering Apple's M4 Max Tensor Path

Apple shipped a tensor matmul in Metal 4.1 and documented almost nothing about how it actually behaves on the M4 Max. So I measured it. Eleven hidden specifications later, the headline is uncomfortable: the fp8 path is emulated, not accelerated — and that turns out to be the most useful thing you can know before you write a single kernel.

June 12, 2026
AI ResearchEvolutionary Mixture-of-LoRA

What If LLMs Had Life? The Strange, Sobering Lessons of UYIR

I spent six months trying to make a language model evolve — to give it a population of adapters that compete, die, reproduce, and pass knowledge to their neighbors. The data did not love the story I wanted to tell. This is what it told me instead, why I am writing it down anyway, and where I want to take this next.

May 14, 2026

More Posts

  • Orion: Programming Apple's Neural Engine for LLM Training (Not Just Inference)May 2026
  • 20 People, 20 GPUs, One Model: What KALAVAI Is Really AboutMay 2026
  • What I Found in the Pentagon's New UFO ReleaseMay 2026
  • Curiosity, Cost, and Coordination: What ACAR Actually Is AboutFeb 2026
[email protected]LinkedInGitHub

Austin, TX