Two LLMs, One GPU, and a Smart Router: Building an Agentic Stack on Kubernetes
Last Updated on June 8, 2026 by Editorial Team
Author(s): Renjith Ravindranathan
Originally published on Towards AI.
A walkthrough of vLLM, the Semantic Router, and Agentgateway — with all the bits the docs leave out.
I’ve been obsessed with one question lately: can I run a real, production-shaped agentic stack on a single homelab box? Not a notebook. A proper Kubernetes cluster, with two specialized models, smart routing between them, and tools the LLM can actually call.
After the introduction, the article walks through designing an agentic stack that uses two different model backends (a coder model and a reviewer model) served by vLLM on a single GPU in Kubernetes, fronted by Agentgateway (Gateway API–compatible proxy) and a Semantic Router that chooses the correct backend by intent. It covers cluster prerequisites, installing Gateway API, deploying Agentgateway and the Semantic Router (including the separate HuggingFace token needed for the embedding model), running two vLLM pods on one GPU via NVIDIA time-slicing while controlling VRAM usage, and mounting a shared HuggingFace model cache to avoid repeated downloads. It then demonstrates routing by headers/intent using Kubernetes route manifests, adding real tool use via MCP (with a federated SQLite MCP example), and testing requests end-to-end through the gateway. The author closes with “homelab-grade” notes on what they’d improve for production (TLS/OIDC, better GPU isolation like MIG, metrics, scaling, disruption budgets), along with wrap-up instructions and references.
Read the full blog for free on Medium.
Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.
Published via Towards AI
Towards AI Academy
We Build Enterprise-Grade AI. We'll Teach You to Master It Too.
15 engineers. 100,000+ students. Towards AI Academy teaches what actually survives production.
Start free — no commitment:
→ AI Engineering For Production — our new book, coming before the end of 2026. Join the waitlist.
→ Agents Architecture Cheatsheet — 3 years of architecture decisions in 6 pages
Our courses:
→ AI Engineering Certification — 90+ lessons from project selection to deployed product. The most comprehensive practical LLM course out there.
→ Agent Engineering Course — Hands on with production agent architectures, memory, routing, and eval frameworks — built from real enterprise engagements.
→ AI for Work — Understand, evaluate, and apply AI for complex work tasks.
Note: Article content contains the views of the contributing authors and not Towards AI.
Related posts
Recent Posts
![[Day 37/100] Build Your Own Agent Memory Layer with Postgres and pgvector](https://miro.medium.com/v2/resize:fit:700/1*-Doos2nmUT3FPZ-Qxbk3Hw.png "[Day 37/100] Build Your Own Agent Memory Layer with Postgres and pgvector")
