Context Researcher, Agent Post-Training

United States Digital Space LLC; San Francisco (CA); On-site; USD 120,000 - 160,000; Full time

Job summary

United States Digital Space LLC is looking for a Context Researcher to scale compute on context in the Agent Post-Training team. This role involves collaborating with researchers and engineers on model training and improving product interfaces.

Your responsibilities will include designing experiments, owning improvements in the post-training stack, and partnering with product teams to enhance model behavior. The ideal candidate will have a strong background in machine learning and be comfortable tackling complex open-ended issues.

Qualifications

• Strong technical fundamentals in machine learning, software engineering, systems, statistics, or a related field.
• Hands-on experience with LLMs, RL, RLHF/RLAIF, post-training.
• Ability to tackle open-ended problems requiring research and engineering execution.
• Comfortable working across research, product, and infrastructure boundaries.
• Desire to train and ship models that are useful for various users.

Responsibilities

• Design and run experiments to improve scaling of compute on context.
• Own improvements to the post-training stack including RL and data pipelines.
• Build environments that expose model failures for improvement.
• Partner with product teams to translate user needs into model enhancements.
• Debug failures in shipped models and formulate solutions.

Skills

Machine learning fundamentals

Software engineering

Systems and statistics

Experience with LLMs

Hands-on RL and evaluation

Job description

United States Digital Space LLC is looking for a Context Researcher to scale compute on context in the Agent Post-Training team. This role involves collaborating with researchers and engineers on model training and improving product interfaces.

Your responsibilities will include designing experiments, owning improvements in the post-training stack, and partnering with product teams to enhance model behavior. The ideal candidate will have a strong background in machine learning and be comfortable tackling complex open-ended issues.