Real-Time Fraud ML Engineer

Sift; Seattle (WA); On-site; USD 150,000 - 210,000; Full time

Job summary

Sift is seeking an experienced Machine Learning Engineer to bridge data science and distributed systems in a production setting. You will build end-to-end ML pipelines, train merchant-specific models, and serve predictions at scale with low latency.

You will work on an automated ML ecosystem that recalibrates models from streaming telemetry data and collaborate across Core Infrastructure, Product, and Data Science teams to translate fraud signals into robust algorithms.

Qualifications

• 4+ years of professional experience building and deploying large-scale ML models in production.
• Proficiency in Java, Scala, and Python for backend and prototyping.
• Hands-on experience with Databricks and big data frameworks such as Spark, Flink, or Hadoop.
• Strong background in statistical modeling and ML algorithms (XGBoost, NN, clustering).
• Experience reasoning about distributed systems and cloud environments (GCP).

Responsibilities

• Design, build, and deploy online ML models including ensemble and DL approaches.
• Engineer time-series features from massive event streams for low latency inference.
• Maintain automated ML training and deployment pipelines with CI/CD.
• Write high-performance code to minimize scoring latency across distributed services.
• Collaborate with Infra, PM, and Data Science to translate fraud signals into models.

Skills

Java

Scala

Python

Databricks

Spark

Bigtable

Kubernetes

Docker

GCP

Tools

Databricks

Spark

Flink

Hadoop

Bigtable

Kafka

Docker

Kubernetes

Job description

Sift is seeking an experienced Machine Learning Engineer to bridge data science and distributed systems in a production setting. You will build end-to-end ML pipelines, train merchant-specific models, and serve predictions at scale with low latency.

You will work on an automated ML ecosystem that recalibrates models from streaming telemetry data and collaborate across Core Infrastructure, Product, and Data Science teams to translate fraud signals into robust algorithms.