IT Infrastructure & Data Systems
Self-built data infrastructure and real operational recovery work — built to understand the mechanism, not just operate someone else's platform.
Self-Hosted Data Platform — Pawsome Provisions
Designed and self-hosted a full PostgreSQL and dbt data platform from scratch — staging through mart layers, validated with dbt test — with no existing playbook to follow. Configured the entire Dockerized environment end to end: network bridges, volume mounts, service dependencies.
Built a full dbt star-schema producing account health scoring and revenue reporting, plus linked Metabase dashboards. The platform also runs an automated correlation engine collapsing 20 raw multi-source signals into a single correctly-attributed event with zero false positives on the simulated dataset, across the runs tested so far — a real exercise in signal quality and system architecture, not just data storage. Sourcing note: this result is against synthetic data I built myself, not an independently validated benchmark against real-world traffic — worth stating plainly rather than letting "zero false positives" imply more than it's earned.
Stack: Python · PostgreSQL · dbt · DuckDB · Metabase · Docker · Anthropic Claude API
Zero-Documentation Recovery — Milestone Technologies
Inherited a high-volume device lifecycle program — imaging, deployment, end-of-life processing — with no existing process documentation, taken over from a complacent incumbent. Built standard operating procedures from scratch through current-state analysis and gap identification.
Sustained a 99% on-time SLA rate across deployments of up to 400 devices daily while building that documentation, and applied a Lean process improvement framework that reduced operational waste by 29% — the operational foundation that justified the contract win in the first place.
See the Actual Work
This page describes the systems. These links are the systems.