The portfolio · October 29, 2026
Projects that come
from real work.
I build tools, operate local AI systems and document the experiments behind them. Here is what you can inspect today, with the next research project clearly marked.
macfleet-dash
A practical fleet-health dashboard for Macs and Linux systems, with optional Windows and AI-engine integrations. The October 20 release adds read-only SparkDash engine cards, token windows and resource history.
The public release contains source and examples. Private computer inventories and operational snapshots stay with the installation.
Local models on the Spark pair
First-party model testing and practical serving work on two NVIDIA DGX Spark systems. The published test history separates hardware, model artifacts, workloads and recorded decisions.
The latest model-history review is dated October 20. Those records describe their own test periods; they are not a live status page.
From a request to a checked artifact
My work combines agent environments, selected local or cloud models, and tools that produce inspectable results. Published notes cover model identity, update interruptions and conflicting file writes.
Hermes and DeepSeek Harness are upstream projects used in the lab, developed by Nous Research and DeepSeek AI respectively. Workman AI Labs documents its own use and integration work.
AI inside an operating business
Content publishing, deals publications, search visibility and an internal operations cockpit. Public case studies explain selected dated results and useful lessons. The implementations, business rules and customer data remain private.