SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Sardine is a leading agentic risk platform for fighting financial crime, offering integrated solutions that unify data across risk teams to stop fraud in real time, prevent AI-driven attacks, and automate fraud and AML operations. The company's platform leverages one of the fastest-growing fraud consortiums, spanning over 6 billion profiled devices, 800 million consumers, and 3 million businesses worldwide. Trusted by leading companies including FIS, GoDaddy, Intuit, Edward Jones, ZoomInfo, and Checkout.com.
As an Applied AI Research Scientist, you will bring expertise in deep learning and foundation models to advance Sardine's fraud detection capabilities. You'll work with one of the richest behavioral datasets in fraud and risk, including device intelligence, behavior biometrics, session telemetry, payment events, and consortium signals.
Key responsibilities include: identifying and scoping foundation model research opportunities, designing rigorous experiments, and executing on the development roadmap. You'll own the evaluation bar for model performance, including offline benchmarks, time- and entity-aware holdouts, calibration, drift monitoring, and comparisons against classical baselines. You'll take models through the full lifecycle from data prep and tokenization through pretraining, fine-tuning, distillation, quantization, and deployment with tight latency budgets. Collaboration is central—you'll partner with Engineering on training infrastructure, GPU efficiency, and production-scale serving, and work with client-facing teams and customers to translate model capabilities into actionable decisions for risk teams. You'll also partner with Legal, Compliance, and model risk teams to build explainability, documentation, and governance required by bank and fintech regulators.
Required qualifications: 4+ years in applied machine learning, quantitative modeling, or ML engineering, including at least one foundation model you pretrained or substantially adapted and deployed to real traffic. You need hands-on self-supervised pretraining experience, practical fine-tuning and adaptation skills, and production experience with model serving, versioning, monitoring, and rollback. Strong Python and SQL skills are essential, along with the ability to prepare very large datasets and self-manage ambiguous applied research projects with clear communication across data science, engineering, product, marketing, and external partners.
Nice-to-have qualifications include background in fraud, AML, payments, credit, or adversarial machine learning; experience building and evaluating LLM-based agents in production; publications or open-source contributions in representation learning or sequence modeling; and experience with model risk management in regulated financial environments.
Sardine maintains a remote-first culture with hubs in the Bay Area, NYC, Austin, Toronto, and São Paulo. The company values performance over hours worked and offers generous compensation in cash and equity, early exercise for all options, flexible paid time off, comprehensive health insurance, 401k/RRSP matching, home office setup stipend, monthly meal and social stipends, annual health and wellness stipend, and annual learning stipend.
