Mark43: Five Years Building 911 Dispatch
Senior Engineer → Director of Engineering. January 2018 – January 2023.
When 911 dispatchers route an emergency, the software they use can't be down. That was the design constraint I worked under for five years at Mark43, building computer-aided dispatch (CAD), records management (JMS), and the geospatial infrastructure that backs both. The customers were police departments, sheriff's offices, and joint dispatch centers across the country, agencies that had been running on systems built in the 1990s. Failure modes weren't "the website is slow." They were "an officer can't be located on the map."
What I led
- Real-time emergency dispatch (CAD): the core operational tool for ~200 agencies
- Near-real-time GPS vehicle and officer tracking, geospatial infrastructure backing CAD and field-mobile clients
- Modern custody and records management (JMS): built the product, the team, and the architecture from zero to first launch in under a year, on a modern stack against a decades-old paper-and-green-screen incumbent
- Cross-cutting platform work: deployment safety, integration with legacy CAD/RMS providers, and the high-availability requirements that come with always-on public safety
The career arc
I joined as a Senior Engineer (IC5) on the CAD platform in 2018. Over the next five years I moved through Task Force Lead (a year leading the integration of a major acquisition's product into our codebase), Engineering Manager for CAD, Senior Engineering Manager for JMS, and finally Director of Engineering. The throughline: every role had me closer to the agencies using the product, not further from them. The best engineering decisions I made were the ones I checked with a dispatcher before shipping. Across those years I stayed hands-on (pair programming was where most of the team-trust got built), and I learned the management calls that don't show up in retrospectives: the right re-slottings, the right terminations, the right time to fly the whole team to a customer site.
What I took away
The biggest lesson from five years building public-safety software is that "reliable" is a vocabulary problem before it's an engineering problem. A consumer product that's 99.9% reliable is excellent. A dispatch product that's 99.9% reliable is unacceptable. 0.1% of the time, in this domain, is a person not making it home. Teams that build for this domain need to internalize that gap, and most of the leadership work is helping them do that without burning out.