Local LLM Workflows

Related project: Local AI & Linux Infrastructure Lab

A lot of this started simply from wanting more control over local AI tooling and understanding what could realistically run outside of cloud platforms.

Most of the experimentation revolves around:

• local LLM hosting
• VS Code integration
• Ollama
• GPU experimentation
• Linux infrastructure
• offline workflows
• automation-assisted development

One thing that became obvious pretty quickly is that the surrounding infrastructure matters just as much as the models themselves.

Things like:

• VRAM limitations
• model loading times
• prompt management
• editor integration
• terminal workflows
• hardware acceleration
• storage performance

all become important very quickly.

There has also been a lot of experimentation around using local AI more like a technical assistant inside existing workflows instead of treating it like a standalone chatbot.

Current areas of interest include:

• development assistance
• infrastructure management
• configuration generation
• debugging support
• documentation workflows
• automation scripting

The broader goal is building practical local-first AI workflows that remain flexible and understandable.