own your tools · keep your data · go a layer deeper than the API
One thesis: run models locally, keep your data yours. No cloud dependency, no data leaving your machine.
| Project | What it is | Where |
|---|---|---|
| AILog | Open-source CLI, AI-assisted logs | PyPI · GitHub |
| DocuMind AI | On-device document assistant | Play Store |
| Personal RAG | Local RAG from scratch - Ollama + ChromaDB, no LangChain | writeup soon |
Not hot takes, not relays. The stuff you can only write if you've actually built it: what running AI locally really looks like - RAG from scratch, inference plumbing, the parts the tutorials skip - and how low-level systems actually work under the hood.
Mostly on X @Zoddiacc.
A decade-plus as a platform engineer at the layers most people never touch - the ones deciding whether a system actually boots, talks to hardware, and holds together under load. That instinct carries straight into how I build AI.
@Zoddiacc