Project Hydra | Humanoid Robot Teleoperator

Welo Global•Santa Clara, CA

8d•Onsite

About The Position

Are you a hands-on technical professional fascinated by the future of embodied AI? We are looking for a Robot Teleoperator and Data Specialist to join our lab in the Bay Area. In this role, you will be at the literal controls of humanoid robots, capturing the high-quality data necessary to train the next generation of machine intelligence. You will teleoperate humanoid robots and robotic arms to perform complex tasks, ensuring the captured data is meticulously labeled and meets the highest quality standards for AI training.

Requirements

Problem-Solving: Proactive in identifying ambiguous scenarios and escalating technical hurdles.
Communication: Strong ability to document and explain data collection decisions for complex movements.
Collaboration: A balanced approach to giving and receiving technical feedback.
Experience building, configuring, or operating humanoid robots (auto-mechanic experience is also highly valued).
Familiarity with Windows, Ubuntu, VSCode, and Basic Git (checking out branches/pulling).
Proficiency in mounting cameras, connecting trackers, and troubleshooting device connectivity.
Basic Python competency (specifically reading error messages) and the ability to parse system logs to identify relevant info.
Strong English vocabulary for describing actions (e.g., push, pull, fold, place).
Advanced spatial awareness to understand how body movements affect camera angles and data quality.

Nice To Haves

Previous experience annotating Vision Language Models (VLM) is a major plus.

Responsibilities

Operate humanoid systems and robot arms to execute specific movements and tasks.
Evaluate recorded videos and datasets to ensure they meet rigorous technical standards.
Hands-on manipulation of robot arms, including hardware assembly and minor repairs when necessary.
Analyze video feeds to identify discrete actions, determining precise start/end timestamps and temporal boundaries.
Categorize objects and describe spatial relationships with extreme precision (e.g., "lower left side of the table" vs. "table").