Phil Rich
Infrastructure & Platform Engineering
Networks, cloud, agentic engineering. Co-founder of Quiet Loon.
Now
Leading an internal pilot using Claude Code across infrastructure and platform teams. Co-founder of Quiet Loon, an AI-powered family admin tool. Running local LLM inference on consumer hardware to explore the boundary between hosted and on-prem agentic workflows.
Selected work
-
Built an asynchronous extraction pipeline that turns inbound messages into structured events, tasks, and reminders, classifies AI failures into transient versus permanent, and retries three times with exponential backoff before surfacing the error to the user.
-
Agentic engineering programme
Ran an 11-week Claude Code pilot across infrastructure operations, saving 194 hours across 74 tasks. Outputs included codifying 1,158 next-gen firewall rules into Terraform with a six-stage GitOps pipeline, plus a governance framework and proposal for formal adoption.
-
Consumer-hardware LLM inference
Characterised the local inference frontier on a 16GB consumer Blackwell GPU across vLLM, llama.cpp, and Ollama: seven quantisation variants of Gemma 4, scored against a 3-task coding rubric. Deployed the batch tier for an ~200-chat categorisation workload.
-
Cloud network architecture
Architected hub-spoke DMZ networks across two Azure regions with per-workload NSG segmentation, BGP dynamic routing over nine multi-site VPN tunnels, and private endpoints for AI services — every change pipeline-gated, zero click-ops.
Background
Multi-cloud infrastructure (AWS, Azure), network engineering, and platform operations, with recent focus on agentic workflows, local LLM inference, and AI governance.