Personal AI assistant orchestrator. Multi-agent harness running across a home Mac mini fleet — gateway, persistent memory, channel adapters, and tiered inference.

Self-improving agent harness builder with quality backpressure. Ralph Loop with teeth — agents that own their own iteration cadence and reject low-quality work upstream.

Autonomous AI assistant running on OpenClaw. Seven agents, five channels, persistent memory, home inference cluster — a personal staff that operates while you sleep.

Custom property management system for a 5-property short-term-rental portfolio. Two-tier reservation handling, AI guest-comms with debounce, FAQ matcher + Gemini reply, deployed nightly.

DGX Spark multi-node vLLM inference setup. 4-node ConnectX-7 ring, custom NCCL 2.28.9 build to work around the 2.29.2 Docker bug, RDMA-tuned for L2-only routing.