Local AI
Local LLM Lab
Match machine hardware to local model and runtime stacks.
- Hardware profile capture and comparison
- Model fit scoring for local workloads
- Ollama, LM Studio, and llama.cpp planning
The sections
Every page stays separate on purpose. That keeps each tool easier to maintain, test, and extend without turning the whole subdomain into one giant application.
Local AI
Match machine hardware to local model and runtime stacks.
Retrieval
Run local retrieval-augmented generation over your own documents with on-device models.
Metadata
Preview, strip, and write image metadata from a dedicated lab page.
Transfer
An upload lane for large photos and videos without email limits.
Strategy
A risk-based game and room server for LAN play and live territory control.
Relief
Turn an image into a heightmap and mesh preview for print or CNC work.
Strategy
A tray and storage system for tile-based games like Rummikub™, designed as a finished product line.
Desktop
A download page for the Markdown Viewer desktop app.
Media
Queue videos, tune watermark placement, and generate ffmpeg overlay commands.
Pantry
A browser-based recipe shelf with pantry matching and cooking notes.
Concepts
A product concept page for the Glory Hole Putters collection.
Why it's split up
Each section has a different job and a different UI. Keeping them in separate folders prevents one feature from swallowing the whole site and makes the deployment easier to understand.
The hub page stays simple and points at the tools that are actually in use.
Contact
If you want a new section, keep it as a sibling folder with its own page and link it back from the hub. That keeps the site organized.