Projects & Work

A financial wellness startup's founding platform, built from zero in 2023 when no SaaS product covered its small-group model: application and join flow, matchmaking, member app, payments, and a Go-based sync and reporting layer over PostgreSQL, InfluxDB, and Grafana. I made the stack and architecture decisions, turned stakeholder requirements into Kanban work, reviewed and merged code, and led a six-person team that included four people I recruited. I worked directly in the data scientist's R/Python code to make the matching algorithms production- and cloud-ready and deployed them on Hetzner with Docker/Dokku. The business was profitable in year one. When it pivoted to individual coaching, I led the migration to lower-cost commercial software; I've been advisory since mid-2024.

Senior full-stack, DevOps, and architecture contributor to a Laravel home-listing application, February–June 2025. I developed listing data, direct image uploads, and agreement/signature handling; configured Laravel Cloud; and automated deployments. I recommended schemas and packages, with decisions made by the team. We completed application development; public release is unconfirmed.

Policy-as-code AI governance for an agency — declarative routines, version-controlled CLAUDE.md set, drift detection, dev-mode guards, pre-push secret scanning.

At DKBinnovative I built the automation a managed-services provider ran across 2,500 to 3,200 Windows, Linux, and Mac endpoints for healthcare and manufacturing clients: cybersecurity and compliance tracking, distributed Windows update and upgrade management, and workstation and service tooling built through daily PowerShell development. Go programs orchestrated that tooling and aggregated results into internal and client-facing reports. I moved delivered scripts to a GitHub source-of-truth model that runs them in memory from reviewed sources, and I led the PrintNightmare response: PowerShell and Group Policy mitigations first, then a centralized print-management rollout that removed traditional print servers from managed environments.

The Chickasaw Nation runs payment and point-of-sale hardware across casinos, convenience stores, gas stations, restaurants, and boutique retail. As Retail Systems Analyst, and one of two people the enterprise relied on for that domain, I designed and implemented enterprise-wide tracking of payment and POS devices and their locations, across hundreds of units, to support PCI compliance. I spent two years as the PCI-DSS subject-matter expert and lead audit preparer, vetted hardware and software vendors, and led stakeholders through selection, planning, and implementation of retail-systems projects.

Cross-platform agent (Linux/Mac/Windows) + server-side message broker over gRPC + mutual TLS. POC replacement for traditional RMM tooling.

Fedora CoreOS immutable reverse proxy at the network edge, terminating SSL on the LAN side for IDS/IPS inspection. Deployed in production.

Event coordination for hosts and guests: invitations, RSVPs, dish signups, dietary restrictions, reminders, and shared-cost contributions. I conceived the service and directed AI agents through implementation, then reviewed, integrated, and validated the application. I own its architecture and live operation across authentication, payments, background jobs, and cloud infrastructure.

A desktop application for designing and playing guitar-amplifier circuits, combining circuit simulation with a neural power-stage model in C++/JUCE. I directed development with Claude Code doing most implementation, guiding architecture, integration, and acceptance. We traced a failed model retrain to flawed training data and reached 0.48% holdout error against a 2% limit, then checked real-time audio performance. Pre-launch.

Framework-agnostic Web Audio API wrapper; modern successor to ember-audio.

Engineering contributor on an interactive sentencing exhibit at the White House South by South Lawn.

Profit calculations for Etsy sellers, using their CSV exports to account for listing, transaction, processing, and advertising fees. I conceived the product and directed AI-assisted implementation, owning the architecture, review, and acceptance criteria. Development is complete; the product is pre-launch.

Research memory that lets an AI agent turn articles, videos, PDFs, and documents into persistent insights with source tracking. I conceived the system and directed its development. PostgreSQL lexical search supports retrieval; the agent handles interpretation and curation, while Python exposes ingestion, persistence, and search tools.