Systems & Work
The systems below reflect how I approach any problem — personal or professional. The thinking is the same. The scale is the only thing that changes.
I maintain a structured personal knowledge system built around governance, classification, and retrieval — because how you organize what you know determines whether you can actually use it. This includes domain-separated knowledge vaults, intake pipelines, classification logic, and semantic search infrastructure. I know what sensitive information is. I know how it should be classified, routed, and protected. That thinking applies in business as much as it does personally.
I run local AI inference on consumer and enterprise hardware — not because it's novel, but because I want to understand the stack before I recommend it to anyone. GPU resource management, model selection, throughput optimization, context window management, multi-agent council design. I test assumptions against real hardware, not documentation. I understand the gap between what a model can do and what it reliably does under real operating conditions.
I build workflow automation systems that connect intake, classification, processing, and output — turning unstructured information into governed, retrievable knowledge. Event-driven pipelines, webhook orchestration, vector database indexing, automated enrichment cycles. The same principles that govern enterprise data pipelines apply at personal scale. I know both, and I know where the assumptions break down at each.
I probe systems out of curiosity — to find where they break, understand why, strengthen them, and make them resilient. This is not a professional habit. It's how I'm wired. The same instinct that drove me to take apart an ERP at Chatleff until it worked differently is what drives me to build and test personal AI infrastructure now.
The technology changes. The method doesn't. Understand it from the inside, find the brittle points, build something that holds.
I document the specific systems I'm currently standing up — governed AI infrastructure, compliance-grade pipelines, multi-agent councils, encrypted knowledge stores. Not concepts. Running code.