About

TechDigestor covers self-hosted AI, local LLMs, and homelab automation — written from a stack that is actually run, not just researched.

Who writes this

Written by Eric Woo

Self-Hosted AI & Automation Engineer

Eric runs his own self-hosted stack: local LLM pipelines on Ollama with dual-model VRAM scheduling on a single 32GB workstation, n8n workflows in Docker, and a TypeScript automation engine that publishes to WordPress on cron. He writes about the systems he actually operates — configs, failure modes, and GPU bills included.

How this site is made (in plain terms)

I’m honest about the process: articles here are drafted with the help of large language models and then reviewed and edited by a human before publishing. The parts that matter — benchmarks, configs, VRAM figures, failure notes — come from hardware I actually operate: a single 32 GB workstation running Ollama with dual-model VRAM scheduling, n8n in Docker, and a TypeScript automation engine that publishes to WordPress on a schedule.

I don’t invent employers, clients, or war stories. If a number appears in a post, it was measured on that setup; if a limitation applies, it’s stated.

What you’ll find here

• Local LLM benchmarks — tokens/sec by model and quantization on real, affordable hardware.
• Honest tool comparisons — self-hosted AI and automation tools, including where they fall short.
• Homelab automation — wiring local models into everyday self-hosted workflows.

Elsewhere

You can also find my writing on dev.to (@ericwoooo_kr).