Senior Data Scientist - Fraud Specialist

What you'll do:

Design, build and evolve Prima’s proprietary predictive technology, in topics such as the prediction of future claims, tactical pricing, renewal probability, fraud detection and customer lifetime value, etc;

Define and support the implementation in each country of the key tools to understand how the business is growing, for example: conversion, retention, emerging profitability. The monitoring tools need to be consistent at a global level and at the same time able to adapt to the different stages of evolution of each country;

Support the countries in planning the pricing and UW strategy and consequent follow up;

Develop a strong critical spirit and creativity that can act as an engine for Prima's growth.

What we're looking for:

Impressive academic achievements in Mathematics, Computer science, Physics, Statistics, Mathematical Engineering or Physical Engineering;

2+ years of work/research experience in analytical fraud detection in the Insurance, Finance or Law Enforcement  domain;

Applied knowledge of network analysis, anomaly detection or risk scoring and segmentation using semi-supervised, self-supervised and human-in-the-loop machine learning techniques;

Strong Python programming skills - especially in the data domain - and fluent use of GIT;

Excellent English level, oral & written;

Genuine, collaborative mindset that inspires and guides others.

Nice-to-have:

Knowledge of Spark, SQL or any other data-oriented programming language and proficiency with standard data wrangling and modeling libraries (Pandas, Sklearn, Keras, XGBboost, LightGBM, etc.);

Knowledge of BI tools (Tableau or similar);

In-depth knowledge of graph databases (Neo4J, AWS Neptune) and analytics (NetworkX, GNN) or explainability applied to anomaly detection, clustering and unbalanced learning techniques;

Knowledge of multimodal models or NLP for feature extraction or document and media classification and analysis;

Full ownership of the model lifecycle, including monitoring of model performance and data or prediction drift.