Lead ML Engineer - Real-Time Fraud Detection

Unity; Mountain View (CA); Presencial; USD 179.000 - 232.000; Jornada completa

Ventajas ofrecidas por este puesto de trabajo

Health insurance

Commute subsidy

Stock ownership

Retirement plan

Vacation days

Parent leave

Office snacks

Mental health support

Employee groups

Assistance program

Training programs

Volunteer match

Descripción de la vacante

Unity’s Offerwall fraud platform seeks a Machine Learning Engineer to own the ML end-to-end, transforming real-time anomaly detection and device risk models into production. You will build a golden fraud dataset, define evaluation metrics, and ensure models and rules align with payout and blocking decisions.

You will collaborate with data engineering, backend, and product partners; modernize training and serving, evaluate SageMaker to Kubeflow/KServe, and monitor health with runbooks and data

Formación

• Production ML experience in real-time systems.
• Experience with fraud detection or anomaly detection.
• Experience with low-latency model serving and feature pipelines.
• Ability to design evaluation methods for evolving fraud patterns.
• Strong Python and SQL; familiarity with AWS SageMaker, Kubeflow or KServe.
• Collaborative, able to work across backend, data, and product teams.

Responsabilidades

• Ship fraud models to production and influence payout decisions.
• Build and maintain a labeled fraud dataset and evaluation metrics.
• Design streaming and batch features feeding online feature store.
• Evolve training and serving stack; evaluate SageMaker to Kubeflow/KServe.
• Investigate fraud spikes and derive new detection signals.
• Monitor model health and participate in on-call/runbooks.

Conocimientos

Production ML

Fraud detection

Real-time serving

Evaluation methods

Python

SQL

CloudML: SageMaker

Herramientas

Redis

Kafka

Airflow

Kubeflow

KServe

SageMaker

Descripción del empleo

Unity’s Offerwall fraud platform seeks a Machine Learning Engineer to own the ML end-to-end, transforming real-time anomaly detection and device risk models into production. You will build a golden fraud dataset, define evaluation metrics, and ensure models and rules align with payout and blocking decisions.

You will collaborate with data engineering, backend, and product partners; modernize training and serving, evaluate SageMaker to Kubeflow/KServe, and monitor health with runbooks and data