<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>H100 on Anish Bista</title><link>https://anishbista.org/tags/h100/</link><description>Recent content in H100 on Anish Bista</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Tue, 10 Nov 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://anishbista.org/tags/h100/index.xml" rel="self" type="application/rss+xml"/><item><title>Building a Multi-Tenant GPU Cloud on Kubernetes: Kamaji for Isolation, Run:ai for GPU Sharing</title><link>https://anishbista.org/blog/hard-multi-tenancy-gpu-tax-kamaji-runai/</link><pubDate>Tue, 10 Nov 2026 00:00:00 +0000</pubDate><guid>https://anishbista.org/blog/hard-multi-tenancy-gpu-tax-kamaji-runai/</guid><description>&lt;p>Let&amp;rsquo;s say you are building a GPU cloud. You buy H100s and sell them to customers: a bank takes 16 GPUs, a hospital takes 8. Before signing, both ask the same question:&lt;/p>
&lt;p>&lt;em>&amp;ldquo;Can the other customers see my workloads?&amp;rdquo;&lt;/em>&lt;/p>
&lt;p>The answer has to be no, so each customer becomes a tenant with its own Kubernetes cluster.&lt;/p>
&lt;p>A month later, the bank comes back with a new problem:&lt;/p>
&lt;p>&lt;em>&amp;ldquo;Our fraud team&amp;rsquo;s GPUs sit idle all night, while our NLP team waits in a queue for the same GPUs. Can they share?&amp;rdquo;&lt;/em>&lt;/p></description></item></channel></rss>