MLOps Engineer

Role Overview

Own the infrastructure, pipelines, and deployment frameworks that power real-time AI models. You will build robust MLOps platforms to automate the ML lifecycle—training, deployment, monitoring, and scaling—while ensuring reliable, cost-effective, low-latency performance in production cloud environments.

Responsibilities

• Design, build, and maintain CI/CD pipelines for ML models to automate the path from research to production
• Deploy and orchestrate ML workflows using Kubernetes, Kubeflow, and/or Apache Airflow (manage complex DAGs and training pipelines)
• Implement model monitoring, logging, and alerting for data drift, concept drift, and performance degradation in real time
• Optimize model serving infrastructure using Triton Inference Server, TorchServe, or TensorFlow Serving to minimize latency and infrastructure costs
• Build and maintain a centralized feature store (e.g., Feast or Tecton) to standardize data definitions across training and serving
• Work with security and compliance teams on data governance, model lineage, and access controls across the ML lifecycle

Requirements

• 3–6 years of software engineering/DevOps/data platform experience; at least 2 years in MLOps in production
• Strong Python and shell scripting skills
• Experience with Docker and Kubernetes (containerization and orchestration)
• Hands-on cloud infrastructure experience, preferably AWS or GCP
• Terraform for Infrastructure as Code (IaC)
• Familiarity with ML tracking/registry tools such as MLflow and Weights & Biases (or cloud-native model registries)
• Solid software engineering practices: Git workflows, unit testing, and automated integration testing

Bonus

• Experience with large-scale distributed training (e.g., Ray, Horovod)
• Experience deploying LLMs using vLLM or Hugging Face TGI