Staff Machine Learning Engineer

Overview

Own the ML platform layer between data science and infrastructure at Tilt, running the managed feature store and building the training and deployment platforms that let data scientists ship without an engineer in the loop.

What you'll do

• Run the managed feature store: build patterns and tooling for defining, backfilling, and serving features so a new model can be trained from existing features and deployed.
• Make the feature store the starting point for model training, not just deployment, so new models are trained on existing features instead of rebuilt pipelines.
• Build and improve automated training and deployment platforms; ML engineers own frameworks, not individual models.
• Turn the steps data scientists repeat (registering features, launching training jobs, promoting a model, setting up monitoring) into self-service agents and skills they run themselves.
• Standardize training and retraining so refits are schedulable, reproducible and backed by feature store data rather than one-off queries.
• Scale batch scoring and low-latency online inference, with autoscaling, resource isolation and cost controls that hold up as the number of models grows.
• Extend monitoring for data drift, feature drift, prediction quality and serving health, with alerts and dashboards data scientists use without escalating to engineering.
• Define standards for model and feature versioning, lineage, testing and documentation, and lead their adoption across data science teams.

What you'll need

• Deep hands-on experience as a platform or software engineer working in ML, with a track record of platforms that served many models in production across multiple teams.
• Experience with feature stores and model deployment platforms, such as Databricks Feature Store, Feast, Tecton or equivalent; time on a managed feature store is especially relevant.
• Strong CI/CD for ML (GitHub Actions, Jenkins, Azure Pipelines or similar), model registries, and infrastructure as code with Terraform.
• Cloud expertise in AWS or Azure for ML deployment, plus Kubernetes, workflow orchestration, and a strong command of batch and real-time serving, autoscaling, and latency and cost trade-offs.
• Experience removing engineers from the deployment loop with self-service tooling that data scientists actually adopted, and working side by side with backend engineers to get deployments live.
• Ability to work with ambiguity: gather requirements from stakeholders without being directed, propose solutions and drive projects to completion.
• Enough depth in model development and common algorithms to be a real technical partner to data scientists.

Nice to have

• Familiarity with PySpark, Databricks and monitoring tools such as Evidently AI.

Details

• Location: Remote - US.
• Tilt is remote-first, with in-person company offsites at least twice a year.
• Team is collaborating across 14 countries and 12 time zones.