Engineering - Applied ML Engineering - Crafter Pipeline [Q2]

About the role Fraud is adversarial machine learning in the wild. Billions of dollars of GMV, real attackers, real merchants on the line. Shopify Payments needs applied scientists who can both invent and engineer: people who are as comfortable sketching system designs as they are training models, and who know when a simple tree beats a fancy transformer. You’ll join the team building the intelligence layer that protects Shopify’s payments stack. That means everything from classic ML to foundation models built for payments. This is not “call an LLM agent and hope it works.” It’s rigorous modeling, careful system design, and shipping production systems at scale. You’ll work across payments fraud, risk, and trust: spotting abuse before it happens, understanding where our products create exposure, and figuring out how quirks of the global payments ecosystem enable bad behavior—then closing those gaps. If you love cutting-edge methods but reach for the simplest tool that solves the problem, you’ll fit right in. Key Responsibilities: - Design, build, and ship end-to-end ML systems for payments fraud & risk: data pipelines, models, and serving layers. - Develop and productionize models across modalities: - Neural sequence models / transformers for transactional and behavioral data - Tree-based models (XGBoost, LightGBM,...) where they’re the right tool. - Architect systems that use models safely and reliably in a high-scale, adversarial environment. - Partner with engineers, product, and risk domain experts to translate messy, real-world fraud patterns into concrete modeling problems. - Own the full lifecycle of your models: problem scoping, feature design, training, evaluation, deployment, monitoring, and iteration. - Analyze the payments and fraud ecosystem to uncover new attack vectors, risk surfaces, and opportunities for protective products. - Balance precision with pragmatism: know when to ship a simple model today vs. invest in a foundation-model-style approach. - Contribute to and leverage Shopify’s internal AI tooling and platforms to move fast without breaking rigor. - Mentor other applied scientists and MLEs on modeling approaches, experimentation, and system design. Qualifications: - Significant experience (Senior / Staff level) as an Applied Scientist / MLE building production ML systems, not just prototypes. - Strong coding skills in a production language (e.g., Python, Ruby, Go, or similar) and comfort doing real software engineering and system design, not just notebooks. - Deep hands-on experience with: - Tree-based models (XGBoost/LightGBM or equivalent) in production. - Neural architectures for sequences (e.g., transformers, RNNs, attention-based models) and embeddings. - Track record of owning end-to-end ML solutions: from data exploration and feature engineering through to deployed, monitored services. - Comfort operating in adversarial domains: fraud, risk, trust & safety, abuse detection, security, or similar; OR strong experience in payments more broadly and a desire to go deep on risk. - A bias toward simplicity and impact: you’re excited when a well-structured baseline or tree model beats something more complex. - Strong product and systems thinking: you can reason about how your models interact with user experience, operations, and the broader payments ecosystem. - Ability to navigate ambiguity, form your own opinions from data, and push projects from “idea” to “shipped” without a detailed playbook. - Clear communication with both deeply technical peers and non-technical stakeholders. Nice to have (not required): - Experience with real-time decisioning systems, streaming features, and low-latency scoring. - Background in financial crime, KYC/KYB, chargebacks, or card network rules. - Prior work on foundation-model-style systems for non-text domains (e.g., transactions, logs, graphs).