My Private AI Workstation

Features

• Local AI model experimentation and prompt engineering
• Secure data handling and privacy-first system design
• Custom automation for development and productivity tasks

Technologies

Linux, Docker, Python, AI frameworks, secure networking

Overview

This project page describes the Private AI Workstation, a self-hosted environment built to explore AI-assisted workflows, secure automation pipelines, and productivity tooling while preserving privacy and safety. Read the full write-up in the blog post on how it came together.

What This Covers

• Running a local AI model (Ollama, llama3.2:3b) entirely offline on older hardware
• Calibrating memory, context window, and generation settings for a resource-limited machine
• A custom system prompt tuned to a specific, useful writing and coding style
• No cloud dependency: everything stays on the physical machine

Why It Matters

It’s the same instinct behind the rest of my project work: understand how a tool actually works under the hood rather than just using it off the shelf, and keep control of my own data while doing it.