Sean Michael Dugan
Senior backend & full-stack engineer

Sean Michael Dugan.

Open to Staff and Senior backend roles.

Based in Austin, TX
Download resume (PDF) ✉ lunacity360@gmail.comLinkedIn (opens in a new tab)GitHub (opens in a new tab)
How I work

How I use AI coding agents

Work comes in from chats, editors and scripts, waits for my approval, and runs on whichever machine fits, with simple jobs routed to local models. Every change comes back with its checks and logs.

monitor-submit · source envelope v1POST /api/work/jobs → awaiting approvalapprove-handoffexact version and digestback to the queue for reviewstatuschanges and checksChats, editors and scriptsWhere work is submitted from.My approvalLimits, merges and deploys.Intake and queueThe durable job record and the dispatch decision.Model routingsimple → local models (Ollama)complex → hosted coding agentsusage limits shift the balanceWorkers on the fleetEach job runs on a machine that fits.Build serverGPU desktopLaptopPatch, checks and logsEvery change comes back with its results.Monitor hub →Status, fleet and pipeline views.
Chats, editors and scripts

Where work is submitted from.

My approval

Limits, merges and deploys.

monitor-submit · source envelope v1
POST /api/work/jobs → awaiting approval
approve-handoff · exact version and digest

Intake and queue

The durable job record and the dispatch decision.

Workers on the fleet

Each job runs on a machine that fits.

Build serverGPU desktopLaptop
Model routing

simple → local models (Ollama)
complex → hosted coding agents
usage limits shift the balance

Patch, checks and logs

Every change comes back with its results.

↺ back to the queue for review and approval

Monitor hub →

Status, fleet and pipeline views.

Architecture of my AI-assisted development setup.See the monitor hub →
Tools and practices
Tools
Codex for code changes, and Cursor. Every repo has an AGENTS.md file with the rules each agent reads first.
Model routing
Simple jobs route to local models through Ollama, and complex ones to hosted coding agents. Usage limits shift work toward local models when a plan runs low.
Cost control
Each job uses the cheapest tool that can do it. Check-only jobs run as a plain script, and hosted CI stays manual-only.
Checks
One verify command decides pass or fail. Full checks run locally in Docker on a dedicated build server.
Approval
I approve every merge, production deploy and billable change, in my own words. Another agent’s say-so doesn’t count.
Monitoring
The monitor hub shows each project’s state, running jobs, failed checks and what’s waiting on me, plus the CPU, GPU and memory use of each machine in the fleet.

Projects built this wayLocal Discovery PlatformGame Server PlatformLocal AI MemoryThis websiteAll projects ↓

Projects

History by year →

Current

Show 9 earlier projects
Experience

Where I’ve worked

Full history →

to

Dun & Bradstreet

Senior Software Engineer, Hoovers / DataVision

Senior Software Engineer (2021 – 2026) · promoted from Software Engineer (2018 – 2021)

Technical lead on DataVision. Built the connector framework, authentication platform and usage metering.

to

FATHOM Water Management

Software Engineer II

Actor-based meter-vendor integrations and high-throughput meter data collection on Akka Cluster.

to

DaVincian Healthcare

Software Engineering Intern → Software Engineer

Publish/subscribe messaging, a financial-screening routing service and a new OAuth2 / JWT security service.

Jan 2012 – Oct 2013

Primus Real Estate

Web Developer / Web Administrator

Maintained the company’s commercial-property website (PHP, HTML, jQuery).

2015

St. Edward’s University

B.S. Computer Science

Austin, TX

Contact

Let’s talk about Staff and Senior backend roles.

lunacity360@gmail.comLinkedIn (opens in a new tab)GitHub (opens in a new tab)

Download resume (PDF)