Arbetsbeskrivning

Data Scientist (machine learning)

Glimra is building insurance solutions for climate-related risks in agriculture. The models are the product — they're what decides whether a farmer gets paid after e.g. a drought and how much, so they have to hold up to an insurer's underwriters and still be something a farmer would actually buy. We're looking for our first data scientist.

What you'd work on

You'd start on the models we take to market, working with our underwriter and our CTO. It's statistical work on weather and agricultural data, and the samples are small, so validating a model is harder than fitting one.

Later the work moves to satellite data and machine learning, and the modelling gets a lot heavier. So it's really two quite different jobs and we want someone who wants both. Someone who's only interested in deep learning probably won't enjoy the first part.

You'd also be explaining your models to people who aren't data scientists, both internally and externally (the insurers we work with).

What we're looking for

• Strong Python, and statistics you know well enough to defend when someone questions the model.
• Experience validating models on small samples, where it's easy to overfit.
• Machine learning on image data that you've built and validated yourself.
• You're comfortable getting your own models running in production.
• You can explain a model, and where it stops working, to someone who doesn't work with data.
• Geospatial or remote sensing experience is a big advantage. We'll teach you the agriculture and the insurance.

How we work

It's greenfield. We're building the platform right now, so whoever joins gets to shape it. We write tests first, we agree the architecture up front, and we're strict about recording decisions so they don't get re-argued in the next review.

We want team players, not rock stars. Direct feedback both ways, a lot of ownership handed over, and people sharing what they're doing as they go instead of going quiet and coming back when it's done.

You'd be the only data scientist, so you'd own the modelling work end to end.

Terms

• Remote / Stockholm
• Full-time. 6-12 project employment to start, and we want it to continue
• Starting asap