Senior Data Scientist (Find My)

Your devices are lost. They have no network connection. Find My locates them anyway, using a crowd-sourced network spanning nearly a billion Apple devices worldwide.

Behind that network is one of the largest and most privacy-sensitive data problems anywhere. Your primary focus will be navigating large-scale spatiotemporal datasets where geography and time intersect. You'll find the signal, model it, and turn it into decisions, without compromising privacy

Description
We work with a high degree of autonomy and own our results end to end. You'll own analyses and models from the first exploratory question through production handoff.

Our roadmap changes. A capability lands, a new feature requirement comes in, a partner team reprioritizes, and the plan from six weeks ago is now the wrong plan. We rewrite it. If you need a stable twelve-month backlog to do your best work, this will frustrate you. If ambiguity reads as opportunity rather than risk, you'll like it here.

We set the priorities and the problems worth solving. How to solve them is largely yours to work out. We're looking for people who take a goal and run with it, and who push back when the goal looks wrong.

Your models don't stay in notebooks. You'll work closely with the engineering teams who build and run our services, handing off prototypes they can scale into production. That means your work has to be reproducible and efficient.

You explain complex work clearly, in writing and in person. You turn patterns into actionable metrics for non-technical stakeholders, make a convincing case with evidence, and change your mind when the evidence says so. Much of our work is settled through clear writing and good conversation, often with teams on other continents.

The work is on-site in Malmö, Sweden. You'll need to be here, or willing to move here. Working language is English, written and spoken.

Preferred Qualifications
Experience from location-heavy domains such as mobility, delivery, logistics, or mapping.
Experience analyzing data too large for one machine. You care about query and algorithm cost, and when a spatial join across billions of observations takes hours, you find out why before asking for a bigger cluster.
Deep Python, used to get real answers. You know when a quick script settles the question and when it needs to become something that lasts, and you move fast in both modes.
Experience working under privacy constraints, such as aggregation thresholds or differential privacy.
Your work hands off cleanly to engineering: reproducible, tested, and documented

Minimum Qualifications

Hands-on data science experience where your work reached production and was used. You've owned it from the first question through handoff, and you've seen a model that looked right offline behave differently on real data and worked out why.
Strong geospatial fundamentals: geometry operations, spatial joins, spatial statistics, and coordinate reference systems. You know why computing area in EPSG:4326 gives the wrong answer. Tools such as GeoPandas, Shapely, or PySAL.
Experience with raw sensor data: cleaning noisy, irregular observations, combining sources, and deriving new signals that hold up on real data.
Proven modeling range, with tools such as scikit-learn or XGBoost for classical predictive work and Prophet or ARIMA for time-series forecasting, with deep learning, such as PyTorch, when the problem calls for it. You also know when the simpler model is the right one.
Experience with spatial databases and distributed spatial processing, such as PostGIS or Apache Sedona, and with discrete global grid systems for hierarchical spatial indexing.