Build a private cluster for the office
Build a private cluster for the office by John Davies
Room 6 · Conference · 15:00-15:50
Summary
Section titled “Summary”A technical talk on building private, local AI systems for regulated organizations using open-weight models. It covers hardware choices, networking, model selection, private data access, and fast document retrieval, helping small teams reduce dependence on frontier model providers while ensuring privacy, governance, and long-term support.
Timeline
Section titled “Timeline”
- Mac MiNi
- Spork
- if Model doesn’t Fit on one Box
- it Con om e[?] , But still slower
- toilscale cable Not Alaways usable
- Best Model
- Still gwen
- RAG
- homies[?] agent sae[?] claude code
- Code readin ≠ code writing