Real-Time Fraud ML Engineer

Compugra Systems Inc; Dallas (TX); On-site; USD 140,000 - 190,000; Full time

Job summary

Compugra Systems Inc. in Dallas, TX is seeking a Machine Learning Engineer to develop and deploy real-time fraud detection solutions. You will build feature pipelines, integrate models with APIs, and support graph-based fraud detection with Neo4j.

The role emphasizes low-latency inference (<250 ms), collaboration with MLOps, and production-grade reliability to scale detection across complex data pipelines.

Qualifications

• Hands-on experience in Python and ML model deployment.
• Experience with APIs, microservices, and production ML systems.
• Knowledge of data pipelines, data engineering, and feature stores.
• Exposure to Google Cloud Platform, Databricks, Data Lake, or Data Warehouse platforms.
• Basic understanding of MLOps, monitoring, and release support.
• Good communication and problem-solving skills.

Responsibilities

• Build and deploy fraud detection services for production use.
• Develop low-latency inference solutions with a target of less than 250 ms.
• Design feature engineering pipelines for ML use cases.
• Integrate ML models with REST APIs and microservices.
• Support graph-based fraud detection using Neo4j.
• Improve scoring performance, reliability, and scalability.
• Work with MLOps teams for releases, monitoring, and production support.
• Support data quality, governance, and operational activities.

Skills

Python

Real-Time Inference

APIs and Microservices

ML Model Deployment

Data Pipelines

GCP Databricks

MLOps

Monitoring

Production Support

Agentic AI

Tools

Neo4j

Graph Databases

Feature Stores

REST APIs

Job description

Compugra Systems Inc. in Dallas, TX is seeking a Machine Learning Engineer to develop and deploy real-time fraud detection solutions. You will build feature pipelines, integrate models with APIs, and support graph-based fraud detection with Neo4j.

The role emphasizes low-latency inference (<250 ms), collaboration with MLOps, and production-grade reliability to scale detection across complex data pipelines.