Home Server

I built a home server to run language models, train image-generation adapters, and host services I use day to day. Proxmox separates the workloads into Linux VMs, with GPU passthrough for inference and training.

I have fine-tuned and hosted Qwen and Mistral models for local inference and agent workflows, and trained LoRA adapters for image generation. Running these workloads locally gives me control over the models, storage, and serving setup.