Member of Technical Staff

About the role

Preference Model is an AI research engineering company building automated machine learning research infrastructure. The team's view is that current frontier models are brittle when applied to real-world ML tasks, and that the key bottleneck is the lack of high quality reinforcement learning training environments. Their first step is to build RL environments that reflect real-world complexity, with diverse tasks and well-designed reward functions. The founding team previously worked on Anthropic's data team, building data infrastructure and the datasets behind Claude. In this role you will help build the infrastructure that powers frontier post-training on large language models. Your work will support high-throughput systems and shape how research is run, bringing the team closer to models that can train themselves. The role is based in San Francisco and is full-time. Frontier research moves only as fast as its infrastructure permits, so the systems you build are not supporting work, they are the foundation of the mission. This is a chance to work directly on the plumbing that decides how quickly new research ideas become real, repeatable capabilities.

Key facts

-

Location: USA

-

Engagement: Full-time.

-

Compensation: in the listing.

• Category: ML Infrastructure Engineering, Research Engineering.
• Focus: Infrastructure for post-training and reinforcement learning on large language models.

What you'll do

• Design and build scalable infrastructure for high-throughput LLM post-training workloads.
• Develop RL environments with diverse tasks and well-designed reward functions.
• Engineer systems that let research teams run experiments reliably and at scale.
• Build data pipelines that feed training and evaluation workflows.
• Optimize infrastructure for throughput, cost, and reliability.
• Operate and monitor the systems you build, responding to issues as they arise.
• Work closely with researchers to understand their needs and translate them into tooling.
• Establish testing and validation practices for infrastructure components.
• Document architectures and operational runbooks for the team.
• Contribute to a small, senior team where everyone owns outcomes.
• Shape the roadmap for infrastructure as the research agenda evolves.
• Review and improve the reliability of existing systems as workloads grow.
• Collaborate with partner labs on shared infrastructure and tooling decisions.

Requirements

• Senior-level experience in machine learning infrastructure, platform engineering, or related fields.
• Strong programming skills in Go and Python.
• Hands-on experience with Kubernetes and containerized workloads.
• Experience running machine learning workloads, including training and evaluation.
• Familiarity with deep learning frameworks such as PyTorch.
• Solid understanding of AWS or GCP cloud environments.
• Experience building data infrastructure and pipelines.
• Ability to work independently in a fast-moving research environment.

Nice to have

• Prior experience on a data or infrastructure team at a leading AI lab.
• Experience with reinforcement learning pipelines.
• Familiarity with large language models and their training requirements.
• Background in running high-throughput distributed workloads.

Skills & tools

• Go
• Kubernetes
• AWS and GCP
• PyTorch
• LLM and ML workloads
• Data engineering and pipelines
• Infrastructure monitoring and operations

Practical notes

• The role is based in San Francisco and is full-time.
• Compensation details are not listed in the posting.
• The company is early stage and the team is senior and small.
• Expect a fast-paced research environment with high ownership.

Project highlights

• Building the RL environments that could unlock self-directed learning for models.
• Creating infrastructure that directly shapes how frontier AI research is run.
• Working with founders who built data infrastructure behind Claude.

Why you should apply

• You will work on the frontier of LLM post-training and reinforcement learning.
• Your infrastructure will directly enable new research capabilities.
• The team is senior, small, and outcome focused.
• You will work closely with researchers at leading AI labs.
• The mission is to push AI toward real-world reliability and self-directed learning.
• You will learn the deepest parts of LLM training infrastructure from people who built it for Claude.