About this role
The Link Fraud and Auth team at Stripe develops a digital wallet designed for fast and secure online payments, alongside fraud protection systems and instant bank payment products. Engineers on this team build and operate machine learning models and risk decisioning systems that secure transactions while maximizing authorization rates for legitimate users across various payment features.
This role involves managing the full machine learning lifecycle, from analyzing fraud patterns and designing data pipelines to deploying, monitoring, and iterating on production models. Professionals who thrive in this environment possess strong technical judgment, enjoy tackling ambiguous problem spaces, and excel at collaborating cross-functionally with product, engineering, and data science partners to scale financial infrastructure.
What you will do
- Build, train, evaluate, deploy, and own machine learning models that detect fraud and abuse across the Link network.
- Use large-scale datasets to investigate emerging threats, develop hypotheses, and identify opportunities to improve payment performance.
- Develop pragmatic machine learning solutions, including tree-based models and other approaches suited to real-time risk decisioning.
- Design data pipelines, features, evaluation methods, experiments, and monitoring systems that support reliable production models.
- Build and improve risk decisioning systems that integrate with other parts of the broader Stripe payments stack.
- Own ambiguous problems from initial analysis and problem definition through technical design, implementation, launch, measurement, and iteration.
- Collaborate with Engineering, Product, Data Science, and Risk partners to turn model improvements into durable product outcomes.
What you need
- Bring at least 6 years of industry experience building and shipping machine learning models in production environments.
- Demonstrate strong programming skills in Python alongside practical experience with common data and machine learning tools such as SQL, Spark, and XGBoost.
- Possess a deep understanding of production machine learning systems, covering data pipelines, feature development, evaluation, deployment, monitoring, and iteration.
- Show documented experience working with large and complex datasets while applying fundamentals in data analysis, statistics, and experimentation.
- Exhibit the ability to take an open-ended business problem, determine where machine learning can help, and own the solution through production.
- Apply strong judgment when selecting practical modeling approaches and evaluating tradeoffs among model performance, system complexity, latency, and business impact.
- Demonstrate strong collaboration skills with the ability to work across teams and contribute to the success of peers.
Nice to have
- Experience applying machine learning to fraud detection, risk modeling, payment authorization, identity, account security, or another adversarial domain.
- Experience building real-time, low-latency machine learning or risk decisioning systems at scale.
- Experience integrating models into production services and designing reliable systems around model outputs.
- Experience with payments, fintech, digital wallets, or money movement.
- Strong software engineering skills and experience designing solutions across the machine learning and product stack.
Skills and tech stack
Questions candidates ask
What team will I be working on?
You will join the Link Fraud and Auth team, which protects consumers and merchants from fraud, abuse, and financial loss while maximizing authorization rates for good users.
What kind of machine learning models will I build?
You will develop pragmatic machine learning solutions, including tree-based models and other approaches suited to real-time risk decisioning and fraud detection.
What is the experience requirement for this role?
Candidates need 6+ years of industry experience building and shipping machine learning models in production.