Senior Data Scientist (Claims Fraud)

About Marshmallow

We exist to make migration easy.

A systemic problem of this magnitude requires a team of curious thinkers who relentlessly pursue solutions. Those who constantly challenge the why, dismantle assumptions, and always take action to build a better way.

A Marshmallow career is built on a cycle of continuous growth, with learning at its core. You will be challenged to raise the bar on your capabilities and supported with the right tools and guidance to do so. This ensures you can deliver impactful work and drive change.

If life at Marshmallow sounds like it could be for you, explore our Culture Handbook to find out more.

Move our mission, and your career, forward.

The Claims Fraud Team

We are here to use data and AI to protect Marshmallow from claims fraud, ensuring that the claims we pay are genuine and fairly assessed. Our work is critical to maintaining a healthy loss ratio, enabling sustainable growth, and building trust with underserved customer segments who rely on us during difficult moments. Our focus is on detecting and preventing fraud at the point of claim. This includes identifying exaggerated or fabricated losses, staged or induced accidents, organised fraud networks, and
inconsistencies in claim narratives or supporting evidence. We build models and tools that help Marshmallow validate claims efficiently and at scale, automate document and image reviews, and direct investigative effort to the claims with the highest fraud risk. The goal is to reduce operational workload while improving decision quality and customer fairness.
You will join a cross-functional team spanning claims operations, product, underwriting, and engineering, with full ownership of your initiatives. You will help shape the next generation of claims fraud defences powered by intelligent workflows, multimodal data, and a strong focus on experimentation and measurable outcomes. Success in this role requires deep technical skill, curiosity about fraud behaviours, and the ability to turn insight into action.

What you’ll be doing

• Building and deploying models to detect claims fraud and suspicious loss patterns
  across FNOL and the full claim lifecycle.

• Embedding risk-based decisioning and triage into our claims platform to ensure that
  the right level of investigation happens at the right time.

• Improving fraud review prioritisation with smarter routing / decisioning logic that
  balances loss ratio impact, operational cost, and customer experience.

• Evaluating the impact of anti-fraud interventions through uplift modelling and
  automated decision tooling.

• Feeding fraud intelligence into model development, working closely with our claims
  investigations, liability, SIU, and analytics teams.

• Contributing to the evolution of our claims fraud infrastructure, shaping how advanced
  modelling, automation, and decisioning are used across multiple Marshmallow claims
  products and processes.

• Using structured and unstructured data to automate fraud checks and evidence
  validation.

Who you are

• You are naturally curious about fraud and interested in how it evolves in the real world

• You care deeply about business impact and think commercially when deciding how to
  apply your skills

• You are pragmatic about delivery and understand when good enough is better than
  perfect

• You are a strong communicator and collaborator across both technical and non-
  technical teams
  

What we're looking for from you

• Experiences that are essential

  • Background in fintech, insurance, or similar industries with exposure to fraud or credit
    risk

  • Experience building and deploying ML models that deliver measurable business
    results

  • Experience working with unstructured data (documents, images)

  • Understanding of how to evaluate performance within risk decisioning, including
    performance trade-offs and operational considerations

  • Proficiency in Python and SQL and confidence working with large datasets

  • At least 3 to 4 years of professional experience in data science or similar roles
    

• Experiences that will help you

  • Experience working on car insurance fraud or understanding of fraud risks in insurance pricing and onboarding

  • Knowledge of industry tools such as Onfido, SIRA, IFB data, or graph-based fraud
    detection platforms

  • Experience deploying fraud solutions into customer-facing products or automated
    decision systems

  • Awareness of how to balance fraud prevention with customer experience and
    operational efficiency

  • Familiarity with document and identity verification processes

  • Experience designing, building and deploying Generative AI applications or Retrieval-
    Augmented Generation (RAG) systems in production environments
    

Perks of the job

• Flexi-office working – Spend 2 days a week with your team in our Budapest office. The rest is up to you! 🏠 Plus you have 4 weeks of Work From Anywhere to use, with no need to come to the office.

• Competitive bonus scheme - designed to reward and recognise high performance 🌟

• 4-week fully paid sabbatical after being with us for 4 years 🌍

• Learning and development – Personal budgets for books and training courses to help you grow in your role. Plus 2 days a year - on us! - to further your skillset 🤓

• Mental wellbeing support – Access therapy and mental health sessions through Oliva 💚

• SZÉP card - Budget to spend on meals, leisure and accommodation 🏝

• Medicover Blue Package - An exclusive pass to top-notch healthcare services 🩺

• All You Can Move membership OR Monthly BKK pass - Unleashing your inner fitness guru or having a hassle-free commute - we've got you covered! 🤝

• Company-wide Marshmallow Engineering Hackathon twice a year 💻

Plus a monthly team social budget, bi-weekly office lunches and office tea, coffee and snacks!

Our process

We break it up into a few stages:

• Initial call with our Talent Acquisition Partner - 40 mins

• A past experience interview where you will discuss your journey so far and ways of working with our hiring manager, also a bit of SQL exercise - 60 mins

• A technical interview with a couple of the team, with code review - 90 mins

• A culture interview with a bar raiser to see if your work style fits our processes and values (and vice versa!) - 60 mins
  

Background checks
As part of our commitment to maintaining a safe and trustworthy environment, we’ll carry out standard background checks, including a DBS and a Cifas check. These help ensure there are no ongoing criminal proceedings and support the prevention of fraud and other forms of serious misconduct. If anything of concern is identified, it may affect your eligibility for certain roles or services. Feel free to ask our Talent Acquisition team if you have any questions about this!

#LI-OK1

Everyone belongs at Marshmallow

At Marshmallow, we want to hire people from all walks of life with the passion and skills needed to help us achieve our company mission. To do that, we're committed to hiring without judgement, prejudice or bias.

We encourage everyone to apply for our open roles. Gender identity, race, ethnicity, sexual orientation, age or background does not affect how we process job applications.

We're working hard to build an inclusive culture that empowers our people to do their best work, have fun and feel that they belong.

Recruitment privacy policy

We take privacy seriously here at Marshmallow. Our Recruitment privacy notice explains how we process and handle your personal data. To find out more please view it here.