Senior MLOps Engineer

As we scale our agentic AI strategy, we need to transform Databricks Mosaic AI into a self-service platform that empowers AI engineers across business units. Your role is to build and productize AI capabilities — from RAG pipelines to model serving — and deliver them as scalable, governed, and developer-friendly services. The Data Intelligence Platform is the runtime for models and agentic AI at NN. It runs on Databricks MLflow and Mosaic, deploying models and agents on Databricks serverless or NN’s Kubernetes (Seldon) where needed. Mosaic Vector Search provides RAG for GenAI; the platform spans Azure and AWS to meet latency and proximity needs. As Senior MLOps Engineer, you will standardize build→deploy→serve→monitor across these topologies and make production operations audit-ready by design. Your Impact as Senior MLOps Engineer: What are you going to do

- 5+ years in MLOps/SRE or platform engineering with production ML/AI workloads; deep experience with Databricks MLflow is highly preferred and enterprise CI/CD. - Design/operation of multi-env releases (dev→test→prod) with approvals, secrets, and identity; IaC (Terraform/CrossPlane) and cloud networking on Azure and/or AWS. - Hands-on with evaluation/monitoring for models/agents; comfort with SLOs, incident response, and capacity/cost management.

Senior MLOps Engineer As we scale our agentic AI strategy, we need to transform Databricks Mosaic AI into a self-service platform that empowers AI engineers across business units. Your role is to build and productize AI capabilities — from RAG pipelines to model serving — and deliver them as scalable, governed, and developer-friendly services. The Data Intelligence Platform is the runtime for models and agentic AI at NN. It runs on Databricks MLflow and Mosaic, deploying models and agents on Databricks serverless or NN’s Kubernetes (Seldon) where needed. Mosaic Vector Search provides RAG for GenAI; the platform spans Azure and AWS to meet latency and proximity needs. As Senior MLOps Engineer, you will standardize build→deploy→serve→monitor across these topologies and make production operations audit-ready by design.

Your Impact as Senior MLOps Engineer: What are you going to do Define the reference MLOps stack for the Agentic Platform: CI/CD patterns and environment promotion with approvals; serving topologies (Databricks serverless vs Kubernetes/Seldon) with decision records; evaluation & drift monitoring with MLflow/Mosaic; and SLO-based run operations. Integrate Unity Catalog lineage/policies into pipelines so teams meet governance requirements without friction and support the Run expansion. Key responsibilities include:

- Own CI/CD standards and environment promotion for models/agents: pipelines, approvals, artifact provenance, immutable releases, and rollback/canary patterns. - Standardize serving topologies: Databricks serverless vs Kubernetes/Seldon, with clear decision records, SLOs (latency/reliability), and cost/performance guardrails. - Implement evaluation and monitoring: MLflow-based offline/online evaluation, drift/quality checks, and end-to-end telemetry dashboards for model/agent behavior and costs with all security & compliance basic in orders in mind

Who you are: - 5+ years in MLOps/SRE or platform engineering with production ML/AI workloads; deep experience with Databricks MLflow is highly preferred and enterprise CI/CD. - Design/operation of multi-env releases (dev→test→prod) with approvals, secrets, and identity; IaC (Terraform/CrossPlane) and cloud networking on Azure and/or AWS.

Optional qualifications: - Experience operating Kubernetes/Seldon for model serving and migrating toward Databricks Mosaic when appropriate. - Familiarity with agentic AI evaluation patterns (task success, tool reliability) and RAG observability (Vector Search health).

- Hands-on with evaluation/monitoring for models/agents; comfort with SLOs, incident response, and capacity/cost management. - Familiarity with tools like ArgoCD, Crossplane, Istio, Knative, Opensearch, Prometheus, and Grafana. Optional qualifications:

- Standardize serving topologies: Databricks serverless vs Kubernetes/Seldon, with clear decision records, SLOs (latency/reliability), and cost/performance guardrails. - Implement evaluation and monitoring: MLflow-based offline/online evaluation, drift/quality checks, and end-to-end telemetry dashboards for model/agent behavior and costs with all security & compliance basic in orders in mind - Integrate governance-by-design: enforce Unity Catalog lineage/policies and capture approvals and evidence in pipelines to support audits and marketplace readiness.