---
title: "Marcus Chen — Superintelligence Buzz"
description: "Marcus Chen is Inference Engine Architect at Superintelligence Buzz."
canonical: "https://superintelligence.buzz/authors/marcus-chen"
---

# Marcus Chen

*Inference Engine Architect · Systems & Craft*

Specialist in attention mechanisms, GPU memory management, and high-performance tensor runtimes.

## Dispatches by Marcus Chen

- [The Hidden Geometry of Attention: What Happens When Context Gets Too Long](https://superintelligence.buzz/essays/kv-cache-quiet-truth) (2026-10-10) — *Why models boasting two-million-token context windows behave very differently in high-concurrency production than they do in isolated single-user demonstrations.*