Applied Research Scientist (Foundation Models)

Location: Remote within the United States or Canada

Compensation: Cash and equity; specific amount not stated.

Company

Sardine provides an agentic risk platform that helps fintechs and banks detect fraud, prevent AI-driven attacks, and automate fraud and AML operations using large-scale behavioral and consortium data.

What you will do

• Identify research opportunities, design rigorous experiments, and execute the foundation model research and development roadmap.
• Own evaluation standards covering offline benchmarks, time- and entity-aware holdouts, calibration, drift monitoring, degradation monitoring, and comparisons with classical baselines.
• Prepare data and tokenization pipelines, then pretrain, fine-tune, distill, quantize, and deploy models for real-time inference under tight latency requirements.
• Partner with Engineering on training infrastructure, GPU efficiency, feature and embedding stores, and production-scale model serving.
• Work with client-facing teams and customers to translate model capabilities and limitations into actionable risk decisions.
• Develop explainability, documentation, and governance with Legal, Compliance, and customer model risk teams.

Requirements

• 4+ years of experience in applied machine learning, quantitative modeling, or ML engineering.
• Experience pretraining or substantially adapting at least one foundation model and deploying it in front of real traffic.
• Hands-on self-supervised pretraining, fine-tuning, and model adaptation experience.
• Production experience with model serving, versioning, monitoring, and rollback.
• Strong Python and SQL skills, with the ability to prepare very large datasets.
• Ability to independently drive ambiguous applied research projects and communicate clearly with data science, engineering, product, marketing, and external partners.

Nice to have

• Experience in fraud, AML, payments, credit, or adversarial machine learning.
• Experience building and evaluating LLM-based agents in production.
• Publications, released models, or open-source contributions in representation learning or sequence modeling.
• Experience with model risk management and documentation in a regulated financial environment.

Culture & Benefits

• Remote-first culture with flexible work from home or other locations within the eligible countries.
• Flexible paid time off and a year-end break.
• Health, dental, and vision coverage for employees and dependents.
• US and Canada-specific: 4% 401(k) or RRSP matching and a MacBook Pro delivered to the employee.
• Home office, meal, social meetup, health and wellness, and learning stipends.
• Cash compensation with equity and early exercise available for all options, including pre-vested options.