ML Infrastructure Engineer: Scale Models in Production

Kurai; Austin (TX); On-site; USD 140,000 - 190,000; Full time

Job summary

Kurai is seeking an experienced MLOps Engineer to design and maintain ML pipelines from research to production. You’ll work across data science and engineering to automate training, feature engineering, and deployment workflows.

Our stack emphasizes Kubeflow, MLflow, and custom tooling to keep latency sub-100ms for millions of predictions daily. You’ll build CI/CD pipelines, operate GPU-accelerated ML platforms on Kubernetes, and implement reproducible ML with feature stores and data versioning.

Qualifications

• Strong Python skills and experience with ML frameworks (PyTorch, TensorFlow, scikit-learn).
• Production experience with Kubernetes, Docker, and GPU orchestration in ML pipelines.
• Experience with CI/CD and data/versioning tools across ML lifecycles.

Responsibilities

• Design and maintain CI/CD pipelines for ML models using GitHub Actions, ArgoCD, and custom tooling
• Build and operate ML platforms on Kubernetes with GPU acceleration
• Implement feature stores (Feast) and data versioning (DVC, Delta Lake) for reproducible ML
• Monitor model performance in production with drift detection, A/B testing, and automated retraining
• Optimize inference latency through model quantization, ONNX, TensorRT, or custom serving solutions
• Manage ML experiment tracking with MLflow or Weights & Biases; ensure reproducibility
• Collaborate with data scientists to productionize research code and establish best practices
• Implement automated testing for data quality, model validation, and pipeline integrity

Skills

Python

ML frameworks

CI/CD pipelines

Education

BS/MS in CS/Engineering or related field

Tools

Kubernetes

Docker

AWS EKS

NVIDIA GPUs

Terraform

CloudFormation

Pulumi

Job description

Kurai is seeking an experienced MLOps Engineer to design and maintain ML pipelines from research to production. You’ll work across data science and engineering to automate training, feature engineering, and deployment workflows.

Our stack emphasizes Kubeflow, MLflow, and custom tooling to keep latency sub-100ms for millions of predictions daily. You’ll build CI/CD pipelines, operate GPU-accelerated ML platforms on Kubernetes, and implement reproducible ML with feature stores and data versioning.