Senior ML Platform Engineer — Production-Grade Pipelines

Datavant; Saint Paul (MN); Remote; USD 224,000 - 280,000; Full time

Job summary

Datavant seeks a Staff ML Platform Engineer to lead the ML Platform team, owning the paved road for model training, deployment, and observability. You will guide architecture for LLM endpoints, ensure safe handling of PHI, and drive self-serve workflows for Data Science teams.

You will mentor engineers, contribute to IaC, and partner across Data Science, App Dev, and Operations to deliver a cohesive platform. Strong AI tooling usage and async communication are essential.

Qualifications

• 10+ years of software engineering experience with 3+ years in enterprise-scale ML platforms.
• Hands-on production experience with Databricks and/or SageMaker, MLflow, and core ML frameworks (PyTorch/TensorFlow).
• Fluency in Java and Python with depth in Apache Spark.
• Terraform, containers, Kubernetes, and GitHub-based CI/CD for ML workloads.

Responsibilities

• Set technical direction for ML training, serving, and observability.
• Evolve paved-road framework for CI/CD, model-workflow scaffolding, and Databricks Asset Bundles.
• Architect LLM-endpoint serving across providers and self-hosted deployments.
• Own standards and tooling for MLflow, model registry, and observability.
• Collaborate with Data & ML Platform teams to present a cohesive platform.
• Mentor engineers and contribute hands-on code and IaC.

Skills

Databricks

SageMaker

MLflow

PyTorch

TensorFlow

Terraform

Kubernetes

CI/CD

LLM serving

Tools

AWS

Spark

Java

Python

GCP

Job description

Datavant seeks a Staff ML Platform Engineer to lead the ML Platform team, owning the paved road for model training, deployment, and observability. You will guide architecture for LLM endpoints, ensure safe handling of PHI, and drive self-serve workflows for Data Science teams.

You will mentor engineers, contribute to IaC, and partner across Data Science, App Dev, and Operations to deliver a cohesive platform. Strong AI tooling usage and async communication are essential.