Fraud Detection Data Scientist – Payments

BNY; Pittsburgh (Allegheny County); On-site; USD 90,000 - 120,000; Full time

Job summary

BNY is seeking a data scientist/data engineer in Pittsburgh to learn the payments data landscape, build ML models to detect fraud, and deploy end-to-end solutions. The role targets a recent college graduate eager to apply data science to real-world issues in payments risk.

You will work on feature engineering, real-time/near-real-time detection, and explainable AI, while collaborating with fraud SMEs and engineering teams to move models from prototype to production.

Qualifications

• Bachelor's degree or equivalent in STEM.
• Foundational ML knowledge and Python coding experience.
• Familiarity with SQL and distributed/cloud data tools is a plus.

Responsibilities

• Learn payments data landscape from transaction, customer, and channel data.
• Develop, train, and validate supervised and unsupervised fraud models.
• Map fraud categories to detection signals and modeling strategies.
• Design and maintain feature pipelines with data engineering support.
• Apply AI techniques (deep learning, embeddings, LLMs) to improve detection.
• Use model interpretability methods to explain flags to stakeholders.
• Track and report model performance and impact to business.
• Take ideas from hypothesis to deployment and monitoring.
• Stay updated on fraud trends and responsible AI practices.
• Grow depth in model science, feature science, and insight science.

Skills

Python

ML fundamentals

Data analysis

Communication skills

Fraud detection interest

Education

Bachelor's degree in STEM or related

Tools

scikit-learn

PyTorch

TensorFlow

pandas

Spark

Job description

BNY is seeking a data scientist/data engineer in Pittsburgh to learn the payments data landscape, build ML models to detect fraud, and deploy end-to-end solutions. The role targets a recent college graduate eager to apply data science to real-world issues in payments risk.

You will work on feature engineering, real-time/near-real-time detection, and explainable AI, while collaborating with fraud SMEs and engineering teams to move models from prototype to production.