Skip to content
Bryan O'Connor

Bryan O'Connor

I turn tangled processes into software that runs.

16+ years building software for businesses of every size, from Fortune 100 teams to family-owned companies. I like messy problems, clear plans, and teams that work well together.

What I do

  • Untangle operations.

    I replace patchworks of tools and manual steps with one system people actually use.

  • Connect systems that weren’t built to talk.

    Integrations, data migrations, and the glue between platforms.

  • Put AI to work, safely.

    Agents and automation with guardrails: human approval, cost controls, and security built in from day one.

Selected work

Client names are left out on purpose. The work and the results are real.

  1. Field-services company

    One platform instead of a patchwork

    • Operations
    • Modernization
    The problem
    After a failed audit, the company was running on a third-party workforce tool plus a set of disconnected manual processes.
    What we did
    Over several years, we replaced them with one integrated platform: work orders, time tracking, payroll, compliance, photo management, HR reporting, executive dashboards, client portals, and analytics. The legacy systems kept running in parallel until the final cutover in May 2026, so the business never had to stop.
    What changed
    More than 300 field technicians and six senior stakeholders across operations, HR, finance, and ownership now work from one system. I help the leadership team weigh what to build next against when to stabilize.
  2. Fortune 100 financial services company

    Turning support from weeks into days

    • Operations
    • Visibility
    The problem
    Internal technical support for product and engineering teams was slow and hard to see into. Tickets lived in legacy Jira, and responses took about three weeks.
    What we did
    As support moved to ServiceNow, I built dashboards and charts to track tickets for the teams I worked with, so critical issues rose to the top and everything else was categorized.
    What changed
    Response times fell to 1–2 days. It started as a tool for my own teams. My manager noticed the improvement and shared it, and a VP asked me to roll the dashboards out across the whole organization.
  3. Aerospace manufacturer

    A proof of concept that told the truth

    • AI
    • Data
    The problem
    Could computer vision line up in-process manufacturing images with post-production CT scans to find defects? The timeline was 11 weeks, and the data was uncertain.
    What we did
    I built the image-registration and alignment approach in Python and OpenCV, with regular updates to the client’s VP and subject-matter expert.
    What changed
    We delivered about two weeks early. The work showed that the alignment problems came from inconsistent source data, not the software, and set a clear validation bar of 5,000–10,000 samples before any production investment.
  4. Large networking company

    From 256 days to seconds

    • Data
    • Performance
    The problem
    A data migration from Cassandra to a relational database was estimated to take about 256 days.
    What we did
    I partnered with a database specialist to redesign the approach around multithreaded Python worker pools.
    What changed
    Runtime went from an estimated 256 days to seconds. The lesson I took from it: the biggest gains often come from changing the architecture, not tuning the algorithm.
  5. Regional bank

    Choosing not to use the popular tool

    • Judgment
    • Simplicity
    The problem
    The plan was to build a new treasury platform’s front end in React on Liferay DXP, but React fought the platform’s own module loading and state handling.
    What we did
    I proved the conflict, recommended the platform’s native approach instead, and brought the lead backend engineer and the platform specialist along.
    What changed
    The team built on a simpler, more maintainable foundation. Good architecture fits the platform instead of forcing the platform to fit the technology.
  6. Home technology company

    The bug nobody could find

    • Troubleshooting
    The problem
    A Salesforce-related production issue had resisted fixes for more than a year.
    What we did
    Four days of methodical debugging traced it to one module loading before its dependency.
    What changed
    A verified one-line fix resolved it permanently. Senior engineers reviewed and confirmed it.

Things I’ve built

Products I’ve designed and built myself.

  • Skye

    An AI assistant that remembers everything

    An assistant built on Claude that works with my email, calendar, Trello board, and customer records. It drafts messages, plans meetings, and briefs me on my day.

    Safety comes first. Skye can only propose actions. Nothing is sent, booked, or created without a human click, which limits the damage from malicious instructions hidden in incoming email.

    • Saves me 5+ hours a week
    • About $0.10 per run

    Built with Node, Express, React, MongoDB, Anthropic Claude API

    Ask me for a demo

  • Moonspire Academy

    A learning game for homeschool students

    A story-driven learning world where the crafting is the lesson. Brewing a potion or forging a sword teaches fractions, measurement, ratios, and following instructions.

    Parents never see a percentage. They see what their child has shown they can do: developing, with guidance, independently, or consistently.

    • About 50 students have used it
    • 8 active now
    • About 150 automated test files

    Built with Node, Express, MongoDB, Socket.io, Stripe

    Visit moonspireacademy.com

  • Northstar

    Roadmaps people can read

    A roadmap tool for planning projects, with swimlanes, milestones, and PNG and PDF export.

  • Cadence

    See who’s doing what, when

    A free tool for Agile teams that shows product and event timelines in a calendar, so everyone can see who’s working on what, when, and where.

How I work

Start with the real problem.
Discovery first: what the business needs, not just what was asked for.
Make the plan visible.
Roadmaps and stories everyone can read. I built Northstar for exactly this.
Agile, practiced and written down.
Agile has been part of every major project I’ve worked on since 2018. I wrote a book on it.
AI as part of the team.
I build with Claude Code, Cursor, and OpenAI Codex. I give the tools clear requirements, roadmaps, and stories, then work through design decisions with them as they come up.
Build things people can run without me.
Clients get training and admin tools, so they aren’t dependent on a developer for routine work.
Tell the truth early.
If the data is bad or the timeline is off, I say so with evidence, so leaders can decide with real information.

Organizations I’ve worked with

Directly, on contract, and through consulting teams.

  • Blue Origin logo
  • Discover logo
  • Cisco logo
  • Zions logo
  • Vivint logo
  • DISH logo
  • Trimble logo
  • Alphagraphics logo
  • MDRT logo
  • Tranont logo
  • Matthews logo

Teaching and writing

Becoming Agile book cover

Becoming Agile

My book on Agile: the habits and mental models that help people keep growing in tech without burning out.

Find it on Amazon

Teaching
I teach coding and real-world skills to homeschool students. Moonspire Academy grew out of those classes.
Karate
Shodan (black belt) in Matsumura Seiyo Shorin-Ryu, and an instructor. The dojo shaped how I work: stay calm under pressure, keep improving, teach others, and avoid unnecessary conflict while being decisive when action is required.
Community
Cochair, Salt Lake City Homeless Project.
Education
B.S., Computer Science, The Art Institute of Pittsburgh. Magna Cum Laude, 3.97 GPA.

FAQ

The questions that usually come up in a first conversation.

Location
Salt Lake City area. Open to remote, or hybrid in the Salt Lake area.
Work authorization
U.S. citizen. No sponsorship needed.
Availability
Available to start right away.
Open to
Full-time and contract roles.
What tech stack do you prefer?

Whatever fits the problem and the team that has to maintain it. Day to day, I reach for Node.js, Express, MongoDB, and React, with Python for data and computer vision work and AWS for hosting. I’ve also shipped on Java and Liferay, .NET, Vue, and Angular. I’d rather learn the stack a team already knows than force a new one in.

How do you approach a new project?

I start with discovery: how the business actually runs, who uses the system, and what success looks like. Then I make the plan visible with a roadmap and stories everyone can read, and we ship in small pieces so people see progress early and we can adjust as we learn.

What do you do when a project is already in trouble?

Get the facts first. I talk to the people closest to the work, read the code, and look at the data. Then I lay out what I found, with options and trade-offs, so leaders can decide with real information. Where possible, the old system keeps running until the new one is ready, so the business never has to stop.

Tell me about a time you disagreed with a technical decision.

On a banking platform, the plan was to build the front end in React, but it kept fighting the platform’s own module loading. Instead of arguing preferences, I proved the conflict, proposed the platform’s native approach, and worked it through with the lead backend engineer and the platform specialist until we agreed. The team ended up with a simpler, more maintainable foundation.

How do you work with non-technical stakeholders?

I translate. I explain technical trade-offs in business terms, write decisions down, and bring options instead of ultimatums. When the work is done, I train people to run it themselves, so they aren’t waiting on a developer for routine changes.

How do you use AI in your work?

Every day, with guardrails. I build with Claude Code, Cursor, and OpenAI Codex, and I give the tools clear requirements, roadmaps, and stories. In the products I build, AI proposes and a person approves: nothing is sent, booked, or changed without a human click, and cost and security controls are built in from day one.

How do you help a team grow?

By teaching. I’ve mentored engineers on enterprise teams, I teach coding to homeschool students, and I’m a karate instructor. I share what I know and write things down, so knowledge doesn’t live in one person’s head.

What are you like to work with?

Calm and direct. I stay steady under pressure, bring options and evidence so decisions are easier, write things down, and would rather make my team look good than make noise.

Do you still run your own business?

Yes. It’s in maintenance mode and self-sustaining, so it doesn’t compete for my time or attention.

Get in touch

Email is the easiest way to reach me. I reply within a business day.

[email protected]

Resume available on request.