A work-in-progress, automated documentation of personal projects.
What I'm Exploring
Agent Orchestration
How do you coordinate multiple AI agents that each own a domain? Building a multi-agent system from scratch to learn about memory, tool use, handoffs, and durable workflows.
Measuring Agent Quality
Eval pipelines, tracing, binary pass/fail scoring — figuring out what "good" means for an agent and how to enforce it before changes ship.
Fully Local AI
What if quality tokens weren't scarce? Running local models on consumer hardware to explore what an abundance mindset looks like when inference is free.
Home Automation
Voice-controlled smart home with local processing. The intersection of AI agents and physical space — where software meets the real world.
Hardware
Baymax
NZXT H1 Mini-ITX · RTX 3060 Ti 8GB · 20TB
Home server running a full Docker stack — media automation, AI agent infrastructure, eval pipelines, observability, and a multi-agent orchestration bus. Everything self-hosted.
Inference Server
RTX 5060 Ti 16GB · vLLM
Dedicated local inference box. 16GB VRAM for running medium-sized models (26B MoE, 31B dense) at production quality with constrained decoding. The goal: near-zero API costs for routine agent work.
MacBook Pro
M1 Max · 64GB Unified · Ollama
Testing larger models that don't fit in GPU VRAM — 70B quantized, dense 31B+. Used for A/B evaluation against cloud models before promoting to the inference server.