Lead Data Scientist - Credit Risk Modeling

Job-ID: c1a4f6d4-21e5-49fb-8ecc-84a6b28cfb20Location: MilanSalary: 65133 - 93142 EURType: Full timePosted: Contact: , briefly At Klarna, we're building an everyday finance network, helping over 120 million consumers across 26 countries save time and money, and worry less about their finances. Working here means taking on problems most companies never get to solve, and being hands-on enough that the interesting part of the work lands with you, not someone else — you'll build with AI, not watch it happen. This is the stretch zone. Come find out what you're capable of.

Se è interessato/a a candidarsi per questo lavoro, la preghiamo di assicurarsi di soddisfare i seguenti requisiti, elencati di seguito.
About the role

Every customer who uses Klarna leaves behind a sequence of transactions, logins, and other events. The Next Generation Modeling Techniques team is building a foundation model that turns that sequence into a single, reusable representation of the customer — a transformer trained specifically on Klarna's own data, built in-house rather than adapted from an existing public model, because the patterns it needs to learn don't show up in general-purpose LLM training sets. That representation feeds Klarna's consumer credit underwriting, where how a customer has behaved becomes a signal for how they're likely to behave next. Getting there means solving a problem most transformer work skips over: how to tokenise a purchase amount, a merchant category, and a timestamp into the same sequence a model can learn from, without losing what makes each of those different. This position sits in a small, high-ownership team, close to the architecture decisions that shape the model. It's about the mechanics of transformers themselves — the algorithms, not the orchestration of an existing one — carried from a research decision through to a system running in production.

What you'll do

You'll train transformer-based models on long, real-world sequences of Klarna's transactional and behavioural events. You'll design tokenisation schemes for numerical, categorical, and temporal features, deciding on vocabulary size, sequence length, and how information compounds across a sequence. You'll own the model lifecycle end to end, from data preparation and training through to serving it in production. You'll translate research decisions — architecture choices, tokenisation trade-offs — into production systems that set the direction for how other Klarna machine learning teams work. You'll evaluate the customer representations the model produces against real underwriting outcomes, not just training-loss metrics.

Who you are

You understand the algorithms behind transformer architectures — not just how to orchestrate a pretrained model — and can reason through architectural tradeoffs from BERT onward. You've designed tokenisation schemes for heterogeneous feature types, such as numerical, categorical, and temporal data, in a sequence model. You've taken a model through its full lifecycle before, from training through to serving it in production. You work with Python, PyTorch, SageMaker, and Airflow day to day. You default to owning problems end to end rather than waiting to be told what's next, and you're comfortable when a problem doesn't have an established playbook yet.

Bonus points for

You've optimised GPU-level performance or written Triton kernels. You have machine learning experience outside deep learning — classical ML, statistics, or other modelling approaches. You've worked with large-scale transactional or financial data before. You've worked in ML infrastructure or MLOps. You hold a PhD you completed while researching model architectures or algorithms.

Things you should know before applying

Working together: we value co-located teams; most teams currently meet in the office 2–3 days per week, and this varies by team and can change over time. Non-obvious backgrounds are welcome. Diversity of skills, perspectives and backgrounds is how we create, innovate, and disrupt like no other. Final compensation will be based on the candidate's qualifications, skills, and experience. Concrete beats comprehensive — what you built, what it did, what it cost. xysqume Curious to learn more about Klarna and what it's like to work here?

#J-18808-Ljbffr