Take a model from a notebook to something a field officer uses on Monday, and stays responsible for it afterwards.
Closes20 Nov 2026
Careers
Build the thing that did not exist here before.
Most of what we work on has no local precedent, which means the method is genuinely open and your judgement genuinely matters. It also means the work is harder than it looks.
Why here
Four honest reasons, and one warning.
We would rather you self-select out now than three months in.
The problems are unsolved locallyCrop types, mineral signatures, languages and record structures here are not in the public training sets. You will be deciding the method, not implementing someone else's.
You will see the consequenceOur clients are a few hours away, not on another continent. You will watch somebody use what you built and find out whether it helped.
The whole stack is in the buildingResearch, data engineering, modelling, interface, field hardware. You can follow a problem across all of it instead of handing it over a wall.
Teaching is part of the jobEvery engagement ends with capability transfer. If you like explaining your work, there is a real place for that here.
The warningEarly-stage engineering in this region involves awkward data, unreliable power, procurement, and work that is occasionally tedious before it is interesting. If you want only the modelling, this is the wrong place.
Open roles
What we are hiring for.
Each listing says what you would actually do in the first ninety days.
How we hire
Four steps, about three weeks.
No unpaid multi-week take-homes and no panel of eight.
Conversation
Thirty minutes about what you have built and what you want next.
Not a screening quiz. You should come out of it knowing whether you want the job.
Practical exercise
A problem close to real work, time-boxed.
Drawn from something we have actually had to solve. Paid if it runs beyond half a day.
Technical conversation
We go through your exercise together.
We are interested in the decisions and the trade-offs, not whether you reached our answer.
Offer conversation
Terms, team, and the first ninety days.
Including an honest account of what will be frustrating about the role.
Questions
Before you apply.
No. We care what you have built and whether you can explain the decisions behind it. A geologist who writes decent Python, a surveyor who has automated their own workflow, or a graduate with a serious side project are all people we want to talk to.
A short call, a practical exercise close to real work and time-boxed to a few hours, a technical conversation about that exercise, and a final conversation about how you like to work. Four steps, usually inside three weeks. We pay for the exercise if it takes longer than half a day.
Mostly in the office, because this work involves field data, hardware and a lot of talking at a whiteboard. Flexibility is normal and arranged honestly rather than through a policy document.
Building systems that someone will use on Monday. Not research for its own sake, and not maintenance of somebody else's platform. You will meet the people who will use what you build, and you will see whether it helped.
Write anyway. Tell us what you have built, what you want to work on and why here rather than elsewhere. We keep strong applications and come back to them when a line opens.