Sr. Data Scientist, Fraud Intelligence

Rakuten Rewards•Toronto, ON

22h•CA$107,957 - CA$157,957•Onsite

About The Position

The Senior Data Scientist, Fraud Intelligence, sits within the Rakuten Rewards Trust & Safety function and is responsible for protecting the platform, its merchant partners, and its members from the full spectrum of fraud and abuse. This role owns the end-to-end lifecycle of fraud detection - from exploratory data analysis and behavioral investigation through to building, deploying, and monitoring production-grade machine learning models that operate in real time. You will work across every dimension of member-facing fraud and abuse, including referral gaming, promo stacking, cashback manipulation, purchase-and-return abuse, account takeover, synthetic identity, affiliate fraud, and coordinated ring behavior. This role is for data scientists who default to AI-first. Using frontier models (Claude, Gemini, GPT-4 class) to drive efficiency is an expectation here, not a perk. We want people who reach for AI before a manual process - and can show how it made them faster, sharper, and more impactful. This is a high-impact, lead-leaning individual contributor role where your models and automation directly reduce financial loss and protect the integrity of the rewards experience for millions of members.

Requirements

Active, demonstrated use of frontier AI models in professional work - able to articulate specific examples where AI accelerated analysis or automated a workflow
Hands-on experience building and deploying fraud, risk, or abuse detection models in production - classification, anomaly detection, or behavioral scoring at scale
Strong SQL & Python skills across feature engineering, model development, pipeline construction, and workflow automation
Proven model testing and validation experience - precision/recall trade-offs, threshold calibration, A/B and champion-challenger experimentation
Experience working with rules engines alongside ML models in a fraud decisioning context
Experience with graph-based or network fraud detection to identify fraud rings or coordinated abuse
Strong communication skills - able to translate fraud signals and model outputs into clear recommendations for nontechnical stakeholders
Familiarity with MLOps practices - model versioning, drift monitoring, and production deployment in a cloud environment
Snowflake or equivalent cloud data warehouse experience
5–7 years of relevant work experience required
Bachelor's Degree in Statistics, Mathematics, Computer Science, Economics, or a related quantitative field required
Background in fraud detection, trust & safety, risk modeling, or abuse prevention required
Experience in e-commerce, fintech, digital rewards, affiliate marketing, or payments platforms required
Snowflake or equivalent cloud data warehouse experience preferred
Familiarity with graph database tooling, such as TigerGraph, Neo4j, or Amazon Neptune, is preferred

Nice To Haves

TigerGraph database tooling is a plus
Snowflake or equivalent cloud data warehouse experience preferred
Familiarity with graph database tooling, such as TigerGraph, Neo4j, or Amazon Neptune, is preferred

Responsibilities

Design and deploy end-to-end fraud detection systems - supervised classification, anomaly detection, and behavioral scoring - across the full member lifecycle from account creation through transaction, redemption, and referral
Identify and model platform-specific abuse patterns, including referral fraud, promo stacking, cashback manipulation, purchase-and-return abuse, account takeover, and coordinated affiliate fraud
Use frontier AI models as a force multiplier - compressing investigation cycles, automating workflows, and surfacing signals faster
Build real-time and near-real-time scoring pipelines that deliver fraud risk decisions at the latency required to intervene before financial exposure is realized
Design model validation and testing frameworks - precision/recall analysis, threshold optimization, A/B testing, and champion-challenger testing - to keep detection accurate as fraud patterns evolve
Manage the interplay between ML models and rules engines, knowing when a hard rule is more appropriate than a probabilistic score
Build automated fraud triage workflows that reduce manual investigation queues and scale team capacity
Own incident response - investigation, root cause analysis, and rapid model or rule adjustments to contain exposure in real time
Develop fraud KPI dashboards and present findings clearly to senior and executive stakeholders
Partner with Product, Engineering, Compliance, and Finance to embed fraud controls proactively
Mentor junior analysts in fraud modeling techniques and investigative thinking