Head of Data Science & ML — Real-Time Risk & Fraud

Trustly; San Francisco (CA); Hybrid; USD 195,000 - 280,000; Full time

Benefits offered by this job

Flexible PTO

Comprehensive benefits

Home office allowance

Internet stipend

401k match

Parental leave

Job summary

Trustly is hiring a hands-on ML leader for their San Francisco/New York hybrid hub. You will build and deploy production-grade models for ACH risk, fraud, and account takeover, while mentoring a team of data scientists and ML engineers.

You’ll own the roadmap, drive experiments, and partner with Risk, Product, and Engineering to deliver measurable business value. In this role you will design low-latency, high-volume inference systems, monitor models in production, and translate complex results

Qualifications

• 5+ years of experience building and deploying machine learning models in production.
• 2+ years of leading a team of data scientists or ML engineers.
• Strong Python skills and advanced SQL experience with large datasets.

Responsibilities

• Build, validate, deploy, and continuously improve ML models for risk and fraud.
• Develop real-time features using transaction, account, device, and behavioral data.
• Design experiments to measure impact of model changes and drive business outcomes.
• Mentor and develop a team of data scientists and ML engineers; own the model roadmap.

Skills

Python

SQL

Production ML

Team leadership

Education

Bachelor’s degree in quantitative discipline

Tools

Pandas

Scikit-learn

XGBoost

LightGBM

CatBoost

Databricks

AWS SageMaker

Google Vertex AI

Job description

Trustly is hiring a hands-on ML leader for their San Francisco/New York hybrid hub. You will build and deploy production-grade models for ACH risk, fraud, and account takeover, while mentoring a team of data scientists and ML engineers.

You’ll own the roadmap, drive experiments, and partner with Risk, Product, and Engineering to deliver measurable business value. In this role you will design low-latency, high-volume inference systems, monitor models in production, and translate complex results