The budget is fixed years ahead
A grant or an appropriation is set long before the work starts and does not flex when a cloud bill does. Architecture has to be predictable enough to budget against, not just cheap on the day it ships.
Nonprofit & Public Sector
Software development for nonprofits and public agencies.
Nonprofits and public agencies run lean teams of specialists on budgets set years in advance, and their software output is a decision somebody acts on. Tarmac builds for those constraints: we took the Utah Avalanche Center's forecasting model off a desktop and into production, twice a day, for the people who decide whether a road stays open.
The context
What makes nonprofit and public sector work different
Three constraints that shape the architecture before a line of code is written.
The team is small and expert
These organizations run lean, and the people in them are specialists in their field rather than in software. A tool that adds a step to their day gets abandoned, however good its output is.
The output is a real decision
A road closure, a danger rating, a public warning. Being right matters more than being fast, so the work has to show where every number came from and be trusted before it is relied on.
How Tarmac helps
One senior team, from research model to running system
Data engineering & analytics
Pipelines, warehouses, and dashboards that turn scattered inputs into one governed answer.
AI & applied ML
Models put into production with the validation and monitoring that make their output trustworthy.
Custom software development
Applications built to be maintained by a small team for a long time.
DevOps & cloud
Infrastructure that runs on a schedule and costs what you planned for.
Proof
Results in the field
“Our goal is not to have a model that is smarter than the human. It is a decision support tool.”
Chad Brackelsberg, Executive Director Emeritus, Utah Avalanche Center
Questions, answered
Nonprofit and public sector development FAQ
Do you work with organizations that do not have an in-house engineering team?
Often. The Utah Avalanche Center came to us with a scientific model and no design, no requirements document, and no software team. We supplied the tech lead, the data engineer, and the analyst, and worked directly with their forecasters to find out what the system needed to be.
Can you put a research model into production?
Yes, and it is some of our favourite work. For the Utah Avalanche Center we took SNOWPACK, a snow cover model from the Swiss Institute for Snow and Avalanche Research that ran on a desktop, and built the ingestion, scheduled compute, and visualization around it so forecasters get results twice a day instead of every few weeks.
How do you keep costs predictable on a fixed budget?
By making the architecture follow the schedule rather than the calendar. Compute that spins up for a run and shuts down afterwards costs what the run costs. We choose platforms partly on how easy they are to forecast a year out, because for these organizations that is as important as the number itself.
How senior is the team?
Senior. Our engineers average 14 years of experience and work in your time zone under the Tarmac 10 quality process. You get people who have shipped this before, not juniors learning on your project.
Sitting on a model that never left the lab?
Tell us what it does and who needs the answer. We’ll bring a senior team that puts it in front of them, on a schedule, on a budget you can plan around.