AI Ops Engineer

KLA•Ann Arbor, MI

24d

About The Position

We seek a highly skilled and passionate Senior AI Ops Engineer to join our team. This role will be pivotal in architecting and delivering the automation layer that enables fast, reproducible, and scalable model development—spanning end-to-end experiment management, model fine-tuning pipelines, and Reinforcement Learning with Human Feedback (RLHF). We encourage you to apply if you’re a systems-minded engineer who loves turning research workflows into reliable production-grade pipelines, setting standards, and mentoring others to raise the bar across the organization.

Requirements

Strong proficiency in Python and experience building robust automation frameworks and production-grade services for ML workloads
Hands-on experience with experiment tracking and model lifecycle tooling (e.g., MLflow, Weights & Biases) and reproducible ML workflows
Practical experience fine-tuning modern deep learning models (e.g., Transformers) and familiarity with parameter-efficient approaches (LoRA/PEFT)
Working knowledge of RLHF concepts and pipelines (preference data, reward models, policy optimization) and how to operationalize human-in-the-loop workflows.
Experience with containerization (Docker), orchestration (Kubernetes), and operating GPU workloads reliably at scale.
Experience with CI/CD, version control (Git), and Infrastructure-as-Code (Terraform/Bicep or equivalent).
Excellent problem-solving skills across distributed systems (training jobs, pipelines, compute infrastructure) and strong communication to partner with research and engineering teams.
Bachelor’s degree in Computer Science, Software Engineering, or related field
5+ years of experience in MLOps/Platform Engineering/DevOps/ML Engineering (or demonstrated equivalent impact), including owning production systems and leading cross-team initiatives

Nice To Haves

Prior experience in a similar industry and/or operating ML platforms with stringent IP/security requirements is a plus.

Responsibilities

Implement and operate experiment tracking, lineage, and reproducibility standards (datasets, code, configs, artifacts, metrics) using MLflow/W&B or equivalents.
Build CI/CD for ML: tests (unit/integration), packaging, reproducibility checks, policy gates, automated deployment and rollback strategies.
Design workflow orchestration for large-scale ML jobs (scheduled runs, triggered retrains, parameter sweeps, gated releases) using tools such as Airflow/Kubeflow/Argo or equivalents.
Architect, build, and own automated pipelines for model training, fine-tuning (e.g., PEFT/LoRA), evaluation, and promotion across environments (dev → staging → production).
Establish standardized training “recipes” (configs, templates, golden paths) to reduce time-to-first-experiment and improve consistency across teams.
Enable and optimize distributed GPU training (throughput, reliability, and cost), including checkpointing, mixed precision, fault tolerance, and spot/preemptible handling where applicable.
Develop evaluation harnesses and automated benchmark suites (quality, safety, latency, and cost) with clear, repeatable reporting to compare runs and releases.