MLOps Engineer

Evlo AI; Raleigh (NC); On-site; USD 130,000 - 170,000; Full time

Job summary

Evlo AI in Raleigh is seeking an experienced MLOps Platform Engineer to own CI/CD pipelines for ML models, reproducible training environments, feature stores, and scalable serving infrastructure.

You will build automated pipelines, monitoring, and infrastructure that let ML engineers move fast without breaking production inference, collaborating with data scientists. This role emphasizes production-grade tooling and cost-aware optimization across cloud environments.

Qualifications

• 3–6 years of experience in MLOps, ML engineering, or platform/DevOps engineering with a production ML focus
• Strong Kubernetes experience in production: deployments, autoscaling, GPU scheduling, and debugging distributed workloads
• Hands-on experience with at least one workflow orchestrator (Airflow, Kubeflow, Argo, Prefect) and experiment tracking (MLflow, W&B)
• Proficient in Python and/or Go; comfortable writing tooling, not just configuring it
• Deep familiarity with cloud ML infrastructure on AWS, GCP, or Azure — including cost management and multi-environment setups
• Solid grounding in ML fundamentals sufficient to reason about training jobs, model versioning, and deployment tradeoffs; BS in CS, engineering, or equivalent practical experience
• Bonus: Experience with GPU optimization (CUDA profiling, Triton/TensorRT), feature stores (Feast, Tecton), model registries and governance, or LLM serving stacks (vLLM, TGI)

Responsibilities

• Build and maintain CI/CD pipelines for ML workflows using GitHub Actions, GitLab CI, or Argo Workflows, including automated testing and validation gates for models and data
• Design and operate model serving infrastructure using Kubernetes, KServe, Triton Inference Server, or cloud-native serving
• Implement end-to-end ML pipeline orchestration with tools like Airflow, Kubeflow, or MLflow, covering data ingestion, training, evaluation, and deployment stages
• Stand up model monitoring and observability: latency metrics, GPU utilization, data drift detection, and automated alerting with Prometheus, Grafana, and custom dashboards
• Manage infrastructure as code for ML platforms using Terraform and Helm, spanning GPU clusters, storage, and networking across cloud environments
• Optimize training and inference costs — batch inference strategies, model quantization, autoscaling policies, and spot-instance training jobs
• Partner with ML engineers and data scientists to productionize experiments, turning notebooks into tested, versioned, deployable pipelines

Skills

MLOps experience

Python and Go

Automation & testing

Education

BS in CS, engineering, or equivalent practical experience

Tools

Kubernetes

Airflow

Kubeflow

Argo

Prefect

MLflow

Weights & Biases

Job description

About The Role

The role focuses on the infrastructure layer that makes ML possible at scale: CI/CD pipelines for models, reproducible training environments, feature stores, and serving infrastructure that stays up under load. This is not a role for someone who dabbles in DevOps — it is core platform engineering for machine learning systems.

About The Role

The role focuses on the infrastructure layer that makes ML possible at scale: CI/CD pipelines for models, reproducible training environments, feature stores, and serving infrastructure that stays up under load. This is not a role for someone who dabbles in DevOps — it is core platform engineering for machine learning systems. The team ships models to production weekly, not quarterly. This role is why that cadence is possible: building the automated pipelines, monitoring, and infrastructure that let ML engineers and data scientists move fast without breaking production inference.

Key Responsibilities

• Build and maintain CI/CD pipelines for ML workflows using GitHub Actions, GitLab CI, or Argo Workflows, including automated testing and validation gates for models and data
• Design and operate model serving infrastructure using Kubernetes, KServe, Triton Inference Server, or cloud-native serving (SageMaker Endpoints, Vertex AI Endpoints)
• Implement end-to-end ML pipeline orchestration with tools like Airflow, Kubeflow, or MLflow, covering data ingestion, training, evaluation, and deployment stages
• Stand up model monitoring and observability: latency metrics, GPU utilization, data drift detection, and automated alerting with Prometheus, Grafana, and custom dashboards
• Manage infrastructure as code for ML platforms using Terraform and Helm, spanning GPU clusters, storage, and networking across cloud environments
• Optimize training and inference costs — batch inference strategies, model quantization, autoscaling policies, and spot-instance training jobs
• Partner with ML engineers and data scientists to productionize experiments, turning notebooks into tested, versioned, deployable pipelines

What We Are Looking For

• 3–6 years of experience in MLOps, ML engineering, or platform/DevOps engineering with a production ML focus
• Strong Kubernetes experience in production: deployments, autoscaling (HPA/KEDA), GPU scheduling, and debugging distributed workloads
• Hands-on experience with at least one workflow orchestrator (Airflow, Kubeflow, Argo, Prefect) and experiment tracking (MLflow, W&B)
• Proficient in Python and/or Go; comfortable writing tooling, not just configuring it
• Deep familiarity with cloud ML infrastructure on AWS, GCP, or Azure — including cost management and multi-environment setups
• Solid grounding in ML fundamentals sufficient to reason about training jobs, model versioning, and deployment tradeoffs; BS in CS, engineering, or equivalent practical experience
• Bonus: Experience with GPU optimization (CUDA profiling, Triton/TensorRT), feature stores (Feast, Tecton), model registries and governance, or LLM serving stacks (vLLM, TGI)