03 SELECTED WORK

Case studies from a decade of platform and reliability work, AI engineering in production, and a self-hosted AI homelab.

• A self-hosted AI and services homelabLocal LLM inference on a 128 GB Mac, plus a dependable Docker services stack on a two-core NAS — self-hosted, backed up, and reachable with no open ports.
• Admin and configuration for an AI knowledge assistantThe control surface that turns a retrieval pipeline into a product a non-engineer can run safely.
• A decade on one campus-scale platformTen years of integrations and releases on a platform that native apps can't afford to have break.
• SecureU: PII remediation at a universityTwo years of PII cleanup at institutional scale that still shape how I treat the context window.