AI Systems Engineer | 11 PyPI Packages | 7 Live Deployments | 36,000+ Tests Python · Rust · TypeScript · FastAPI · Sui Move · Full CI/CD
I spent 17 years building reliable systems in manufacturing (IBM, Toyota) — then applied that discipline to AI engineering. The result: production systems with real users, real billing, and real test coverage. Not demos. Not prototypes. Shipped software.
Portfolio · LinkedIn · Substack · Portland, OR
Animus — Multi-Agent Orchestration Framework
Coordinate specialized AI agents with budget controls, quality gates, and checkpoint/resume. Self-improvement loop with Ollama. MCP server for Claude Code integration. FastAPI · React · 4 packages (Core/Forge/Quorum/Bootstrap) · 37 tools · 14,748 tests · 97% coverage · Quorum on PyPI
BenchGoblins — AI Fantasy Sports Decision Engine (Live)
Scored LLM routing (Grok-primary, Claude fallback), 14 commissioner tools, Goblin Verdict scoring. Stripe billing. Dual-token auth. Next.js · FastAPI · PostgreSQL · Redis · 2,873 tests · 98% coverage · benchgoblins.com
EVE Gatekeeper — Route Intelligence Platform (Live)
14-layer interactive map, gate camp warnings, expandable risk breakdowns per hop, SSO auth, Stripe billing. Public repo. Next.js · FastAPI · SQLite · 3,296 tests · edengk.com · GitHub
DOSSIER — Document Intelligence Platform
Investigative-grade analysis: ingest PDFs/emails/scans, extract 4 NER entity types, surface co-occurrence relationships. Zero cloud dependencies. SQLite FTS5 · Custom NER · Auto-classification · 1,084 tests · 97% coverage
Argus Overview — Linux Multi-Window Manager for EVE Online
Real-time 30 FPS window previews, team management, auto-tiling. Wayland support. Cross-platform. 26,000+ downloads · 36 releases · 2,184 tests · 100% coverage · PyPI
WatchTower / Witness — On-Chain Intelligence & Dossier NFTs (Live)
16 Discord commands, Dossier NFT minting (3 tiers), Oracle Loop for entity profiling, reputation cache, Stripe + SUI payments. FastAPI · Sui Move · Discord.py · dApp Kit · 817 tests
Monolith — Anomaly Detection Engine (Live)
35 detection rules, 17 checkers, chain state via Sui GraphQL, public API v1, webhook subscriptions, Canvas2D heatmap with 24K systems. FastAPI · SQLite · Sui GraphQL · 296 tests · 80% coverage gate
Frontier Tribe OS — Corp Operations Dashboard (Live)
Kill Feed, LLM Briefing, System Intelligence, Alert Config, Pilot/Corp Intel, Battle Reports, Signature Resolution. Next.js · FastAPI · 7 modules · Role-gated access
Published Developer Tools — PyPI
| Tool | What | Tests | Install |
|---|---|---|---|
| anchormd | AI agent context file generator/auditor + tech debt scanner · anchormd.dev | 693 | pip install anchormd |
| agent-lint | Workflow YAML cost estimator + anti-pattern linter | 565 | pip install agentlinter |
| context-hygiene | Context window bloat detection + signal density scoring | 398 | pip install context-hygiene |
| promptctl | Claude API toolkit — prompt engineering, code review, doc intelligence | 330 | pip install promptctlai |
| memboot | Zero-infra persistent memory for LLMs — SQLite + TF-IDF | 292 | pip install memboot |
| ai-spend | AI API cost aggregator — Anthropic + OpenAI in one view | 267 | pip install ai-spend |
| mcp-manager | MCP server manager across agentic IDEs | 159 | pip install arete-mcp |
| convergentAI | Multi-agent conflict resolution — Python + Rust (PyO3) | 900 | pip install convergentAI |
| G13-Linux | Logitech G13 Linux driver — PyQt6 GUI, WebSocket API | 1,641 | pip install G13-Linux |
| argus-overview | EVE Online multi-window manager | 2,184 | pip install argus-overview |
| LikX | Screenshot + annotation + OCR for Linux | 1,752 | pip install LikX |
80+ CI/CD workflows · 0 code scanning alerts · 0 dependabot alerts
| Deployments | 9 services on Fly.io + Vercel — BenchGoblins, Gatekeeper, WatchTower, Monolith, Tribe OS, anchormd, AI Cards, StayCards, license server |
| Containers | Multi-stage Dockerfiles, docker-compose, multi-arch builds (amd64 + arm64) |
| Orchestration | Kubernetes (Deployments, HPA, Ingress), Helm charts, Kustomize |
| Security | CodeQL + gitleaks + Bandit + Semgrep + Trivy + pip-audit across all repos |
| Publishing | 11 PyPI packages via OIDC trusted publishing (zero stored tokens) |
| Billing | Stripe integration (payments, webhooks, license key generation) on 3 products |
| Tests | 36,000+ verified across active repos |
| PyPI packages | 11 published, 2 with 16+ releases |
| Live systems | 9 production deployments with real users |
| Argus downloads | 26,000+ (36 releases on PyPI) |
| CI workflows | 80+ across 20 repositories |
| Security alerts | 0 code scanning, 0 dependabot — fleet-wide |
| Domain | Tools |
|---|---|
| Languages | Python, Rust, TypeScript |
| AI/ML | Claude API, OpenAI, Grok, Ollama, multi-agent orchestration |
| Backend | FastAPI, SQLAlchemy, PostgreSQL, SQLite, Redis |
| Frontend | React, Next.js, React Native/Expo, PySide6 |
| Blockchain | Sui Move, GraphQL, dApp Kit |
| Infrastructure | Docker, Kubernetes, Helm, GitHub Actions, Fly.io, Vercel |
| Security | CodeQL, Bandit, Semgrep, Trivy, gitleaks, SAST pipelines |
I scaled an ice cream line from 740 pints/day to 4,800/hour using Kaizen. I apply the same thinking to software: standard work, visual management, poka-yoke, jidoka. Operational reliability over algorithmic elegance. A system that ships predictably beats one that demos brilliantly.