MLOps Engineer

Evlo AI; Los Angeles (CA); On-site; USD 150,000 - 210,000; Full time

Job summary

Evlo AI in Los Angeles is seeking a seasoned MLOps engineer to own the production pipeline that moves models from research notebooks into scalable, reliable services. You will build and maintain end-to-end pipelines, CI/CD for ML, and robust serving architectures designed to handle real traffic.

You will collaborate with ML engineers and data scientists to ensure GPU utilization is optimized, deployments rollback cleanly, and costs stay under control while monitoring model performance and data

Qualifications

• 3–7 years of MLOps, ML engineering, DevOps for ML, or backend engineering with heavy ML infra exposure
• Strong Python and SQL skills, plus fluency in at least one compiled or systems language (Go, Rust, or C++)
• Deep hands-on experience with Kubernetes, Docker, and at least one major cloud platform (GCP, AWS, or Azure)
• Production experience with ML deployment and serving patterns: batch, real-time, streaming, or LLM inference
• Experience building or operating data/feature pipelines with tools like Airflow, dbt, Spark, or Kafka
• Solid grasp of ML fundamentals enough to reason about models, metrics, and failure modes
• Bonus: LLM serving and optimization, infrastructure-as-code (Terraform), or fine-tuning pipeline orchestration; BS/MS in Computer Science or related field

Responsibilities

• Build and maintain end-to-end MLOps pipelines covering data ingestion, feature engineering, training, evaluation, and deployment using tools like Airflow, Kubeflow, or Vertex AI Pipelines
• Design CI/CD workflows for ML systems - automated testing for models and data, staged rollouts, canary deployments, and one-click rollback mechanisms
• Deploy and scale model serving infrastructure using Docker, Kubernetes, and frameworks like TorchServe, KServe, or Triton Inference Server
• Implement production monitoring for model performance, data drift, and system health using tools like Prometheus, Grafana, Evidently, or Arize
• Manage GPU infrastructure and training workloads - optimize utilization, batching strategies, and distributed training setups
• Establish model versioning and experiment tracking standards using MLflow, Weights & Biases, or equivalent tools across the ML organization
• Partner with data scientists and ML engineers to debug production issues, harden pipelines, and reduce inference latency and infrastructure costs

Skills

Python

SQL

Go/Rust/C++

Kubernetes

Docker

Cloud platforms (GCP/AWS/Azure)

ML fundamentals

Education

BS/MS in CS or related field

Tools

Airflow

Kubeflow

Vertex AI Pipelines

TorchServe

KServe

Triton Inference Server

MLflow

Weighs & Biases

Job description

About The Role

The role owns the infrastructure and pipelines that get machine learning models from research notebooks into reliable, scalable production systems. This means building CI/CD for ML, designing serving architectures that handle real traffic, and making sure training, deployment, and monitoring all run without someone manually babysitting a terminal.

About The Role

The role owns the infrastructure and pipelines that get machine learning models from research notebooks into reliable, scalable production systems. This means building CI/CD for ML, designing serving architectures that handle real traffic, and making sure training, deployment, and monitoring all run without someone manually babysitting a terminal. The team sits at the intersection of data science and platform engineering - working directly with ML engineers and data scientists whose models depend on the reliability of what gets built here. The problems are concrete: GPU utilization, feature pipelines that don't silently break, deployments that roll back cleanly, and costs that don't spiral out of control.

Key Responsibilities

• Build and maintain end-to-end MLOps pipelines covering data ingestion, feature engineering, training, evaluation, and deployment using tools like Airflow, Kubeflow, or Vertex AI Pipelines
• Design CI/CD workflows for ML systems - automated testing for models and data, staged rollouts, canary deployments, and one-click rollback mechanisms
• Deploy and scale model serving infrastructure using Docker, Kubernetes, and frameworks like TorchServe, KServe, or Triton Inference Server
• Implement production monitoring for model performance, data drift, and system health using tools like Prometheus, Grafana, Evidently, or Arize
• Manage GPU infrastructure and training workloads - optimize utilization, batching strategies, and distributed training setups (multi-node, multi-GPU)
• Establish model versioning and experiment tracking standards using MLflow, Weights & Biases, or equivalent tools across the ML organization
• Partner with data scientists and ML engineers to debug production issues, harden pipelines, and reduce inference latency and infrastructure costs

What We Are Looking For

• 3–7 years of experience in MLOps, ML engineering, DevOps for ML, or backend engineering with heavy ML infrastructure exposure
• Strong Python and SQL skills, plus fluency in at least one compiled or systems language (Go, Rust, or C++)
• Deep hands-on experience with Kubernetes, Docker, and at least one major cloud platform (GCP, AWS, or Azure)
• Production experience with ML deployment and serving patterns: batch, real-time, streaming, or LLM inference
• Experience building or operating data/feature pipelines with tools like Airflow, dbt, Spark, or Kafka
• Solid grasp of ML fundamentals enough to reason about models, metrics, and failure modes - not just infrastructure
• Bonus: Experience with LLM serving and optimization (vLLM, TensorRT-LLM, quantization), infrastructure-as-code (Terraform), or fine-tuning pipeline orchestration; BS/MS in Computer Science or related field