Remote job
Research Engineer, Universes
Job details
About this role
Role overview
Design and build realistic training environments that help AI systems learn to perform difficult, long-horizon tasks involving ambiguity, interruptions, extended context, and judgment. This research engineering role combines reinforcement-learning research with production-quality software development, including the creation of environments and evaluations that measure genuine capability. You will help move novel methods from experimentation into dependable training infrastructure.
Responsibilities
- Create the next generation of environments for training capable and safe agentic systems. - Design rigorous evaluations that distinguish real capability from superficial task performance. - Collaborate with research and infrastructure teams to integrate environments into production training workflows. - Debug, test, and iterate across both research code and production machine-learning systems. - Contribute to research direction through technical discussion, judgment, and collaborative problem-solving. - Balance exploratory research with robust implementation and delivery.
Requirements
- Strong software engineering skills and the ability to build reliable infrastructure. - High agency, sound technical judgment, and a focus on measurable outcomes. - Ability to operate effectively amid uncertainty and rapidly changing research priorities. - Capacity to combine research exploration with practical engineering execution. - Commitment to developing safe and beneficial AI systems. - Enjoyment of close collaboration, including pair programming.
Nice to have
- Industry experience training, fine-tuning, or evaluating large language models. - Experience with reinforcement-learning environments, simulation, or large-scale ML infrastructure. - Expertise in sandboxing, containers, virtual machines, or distributed systems. - Senior technical experience in a related field or influential research in machine learning.
Benefits and work setup
The position is remote-friendly but travel-required, with locations listed in San Francisco, Seattle, and New York City. Staff are generally expected to work from an office at least 25% of the time, and visa sponsorship may be available depending on the circumstances.