Machine Learning Engineer

In this Machine Learning Engineer position at Watney, you will help improve models by working directly from a live fleet of robots, using real-world video and data to drive training experiments and measure next-deployment performance.

About the Role

At Watney, machine learning engineers convert outputs from an operational robot fleet into better models. The fleet generates substantial video from real work in the field, and turning that material into models that perform better on the next deployment is a core challenge. Model development spans multiple task types, including what the robot perceives, how it decides, and how it acts.

Your work will support the roadmap from imitation learning toward broader generalization. This includes running training experiments, curating and labeling the data that powers them, and evaluating which approaches translate successfully to performance on the robot fleet.

Responsibilities

• Run and evaluate training experiments as models scale.
• Curate and clean the data used to train models.
• Contribute to how new data is labeled and structured.
• Track the metrics that distinguish models that help from models that do not.
• Partner with Teleoperations to understand and improve data at its source.

Requirements

• Experience training models on real-world data pulled from actual operation.
• Experience with imitation learning, reinforcement learning, or a similar method for control.
• Ability to write production machine learning code in Python, PyTorch, or comparable tooling.
• Comfort with cleaning and curating messy, real-world data.

Technologies

• Python
• PyTorch

Mission

Watney’s mission is to expand human ambition in the physical world.

Location

San Francisco, CA (onsite)