About

By day, I'm David Allada - An AI Software Engineer at Lumos, where we build some of the best AI systems in the industry.

By night, I become ~~BATMAN~~ still David Allada - A passionate homelabber, AI enthusiast, and all-around nerd.

I started my homelab a few years ago, inspired by LinusTechTips on youtube. What began with just an old laptop has turned into a small home datacenter. Along the way, I have learned a TON about infra, orchestration, networking, kubernetes, docker, running my own AI models, and so much more! As part of my job, I have been building fully productionized Agents since 2023, when LLama2 was all the craze, and we had to work within a 4096 token context window.

This blog is meant to be part journal, part homelab documentation, and part creative outlet. I want to share my passion and experiments with those who are interested!

What I write about

hosting ai modelsRunning small, medium, and large models on my 6 GPU's. I will share my experiences with how to run various models and my opinions on which are the best.

homelabSix nodes, 110 CPU Cores, 552 GB RAM, 6 GPU's, 400 TB of hard disk storage, and 60 TB of NVMe storage. Workloads between kubernetes (k3s), docker, and virtual machines.

agent harnessesReviewing various agent harnesses and customizing them extensively. Currently running Deepseek Harness, Claude Code, and Pi.

langgraphMy day job - I have used pretty much every feature in LangGraph, particularly Deep Agents.

Get in touch

Shoot me a message on LinkedIn. I read everything, and I try to replyin a timely manner. If you don't hear back, ping me again. Feel free to ask questions about the blog, a specific post, my homelab, or anything in general.

David Allada davidallada