Machine Learning Engineer at Sardine in North America

Job description

Who we are: Sardine is the leading agentic risk platform for fighting financial crime. Our integrated solution unifies data across risk teams to help organizations stop fraud in real time, prevent AI-driven attacks, and automate fraud and AML operations. Sardine’s platform is strengthened by one of the fastest-growing fraud consortiums in the market, spanning more than 6 billion profiled devices, 800 million consumers, and 3 million businesses worldwide. Leading companies including FIS, GoDaddy, Intuit, Edward Jones, ZoomInfo, and Checkout.com rely on Sardine to secure and grow trust in their products. Our culture: • We hire talented, self-motivated individuals with extreme ownership and high growth orientation. • We value performance and not hours worked. We believe you shouldn't have to miss your family dinner, your kid's school play, friends get-together, or doctor's appointments for the sake of adhering to an arbitrary work schedule. • We're a remote-first team spread across time zones, so no office to report to - work from wherever helps you do your best work. Just a couple of things to keep in mind: pay is based on where you're located, and you'll need to keep a home base in the country you're hired in. So while we love the "coffee shop today, mountains tomorrow" life, this isn't a passport-optional, work-from-anywhere-on-Earth kind of remote - think flexible within your country, not borderless. Location: • Remote - United States or Canada About the role: As a Machine Learning Engineer at Sardine, you'll own the systems that make real-time fraud detection possible. Our data science team builds custom models for our clients, you build and run the platform they deploy onto, and the low-latency serving path those models score on. Sardine scores millions of sessions in real time from hundreds of device and behavioural signals, inside a sub-250ms budget. That constraint shapes everything: how features are computed and served, how models are deployed and rolled back, how quickly you know when something has degraded. You'll be the person who figures out why a model broke. What you'll be doing: • Build and own the model serving infrastructure, real-time inference, feature retrieval, and the latency budget that governs both • Build the deployment path our data scientists use to ship models themselves, including bring-your-own-model support for clients hosting their own • Own models in production: monitoring, drift detection, retraining, incident response, and the on-call rotation • Build and optimise the pipelines that turn raw device and behavioural signals into production-ready features • Work across Python and our Go backend to keep inference fast inside the request path • Build models yourself where it makes sense, roughly 20% of the role, and more if you want it • Champion testing, observability, security and compliance in a regulated environment What you'll need • Experience building, not just using, model serving infrastructure. • Production ownership of ML systems: you've been paged when something broke, you found out why, and you changed something so it didn't happen again. • Strong Python, and solid software engineering fundamentals, testing, code review, CI/CD, the discipline that makes a platform other people can rely on. • Comfort with Kubernetes, containers and a major cloud (we're mostly GCP), plus infrastructure-as-code. • Enough understanding of models to debug them. You don't need to have trained one recently, but when precision drops you should know the difference between a data problem, a feature pipeline problem, and a model problem • Experience building tooling other engineers or data scientists actually use, and the judgement to know what should be self-serve and what shouldn't. Bonus Points • Domain knowledge in fraud, risk, or cybersecurity. • Familiarity with CI/CD, Docker, Kubernetes and the modern devops framework. • Understanding of modern browser APIs and high-entropy data collection techniques. • Fa