How to apply for Research Engineer/Research Scientist, Pre-training

About Anthropic

Anthropic is a frontier AI research and product company working on alignment, policy, and security. Its stated mission is to create reliable, interpretable, and steerable AI systems that are safe and beneficial for users and society. The company explicitly notes it does not necessarily recommend working at other positions at Anthropic, so evaluate this specific pre-training role on its own merits.

About the role

This role sits on the Pre-training team, which builds the next generation of large language models. You would work across model architecture, algorithms, data processing, and optimizer development, while also scaling training infrastructure and improving dev tooling. The position combines independent research leadership on small projects with collaboration on larger initiatives, spanning low-level optimizations to high-level model design.

A typical day

A typical day likely involves running or monitoring training experiments, analyzing results, and writing up findings for the team. You may spend time on infrastructure or tooling improvements that make training more efficient or reliable, and coordinate with collaborators on larger pre-training initiatives. The job description does not specify daily routines, so ask about team cadence and remote collaboration practices during interviews.

Who Anthropic is looking for

• Has hands-on experience training or scaling large language models, including work on architecture, optimizers, or data pipelines.
• Can design, run, and analyze scientific experiments independently, and has led at least small research projects end to end.
• Is comfortable contributing across the stack, from low-level performance and infrastructure work to high-level model design decisions.
• Has a track record of improving training efficiency, reliability, or developer tooling in a research or production ML setting.

Tips for this application

• Read Anthropic's career review concerns linked in the job description before applying, and be ready to explain in your materials why you want to work at a frontier AI lab despite those concerns.
• Tailor your resume to pre-training specifically: highlight LLM training runs, architecture or optimizer changes, data processing work, and infrastructure scaling rather than generic ML projects.
• Name concrete results from past training experiments, such as model sizes, compute budgets, throughput or efficiency gains, and what you learned from the outcomes.
• Show evidence of independent research leadership, since the role explicitly involves leading small research projects, not just executing assigned tasks.
• Since the role is remote in the US, address how you collaborate asynchronously and how you have worked effectively with distributed research or engineering teams.

What to cover in your cover letter

Explain why pre-training specifically interests you at Anthropic, referencing the company's stated mission around reliable, interpretable, and steerable AI. Describe one or two research projects you led independently, including the experimental design and what you concluded. Detail your experience with model architecture, algorithms, data processing, or optimizer development at LLM scale. Address how you would contribute to both training infrastructure efficiency and dev tooling, since the role spans the full stack.

Research before applying

• Read Anthropic's published research and any public writing on pre-training, scaling, and alignment to understand the technical direction of the team.
• Review the 80,000 Hours career review linked in the job description on concerns about working at a frontier AI company, and form your own view.
• Check Anthropic's mission statement and recent announcements to understand how pre-training work connects to reliability, interpretability, and steerability goals.
• Look into the backgrounds of Anthropic's pre-training or research teams through public talks, papers, or interviews to understand how the team operates.

Likely interview topics

Based on the job description, expect questions about:

• How you would design and analyze an experiment to test a change in model architecture, optimizer, or data processing for a large language model.
• Specific technical tradeoffs you have made when scaling training infrastructure for efficiency and reliability.
• A research project you led independently: the hypothesis, method, results, and what you would do differently.
• How you think about safety, alignment, and steerability in the context of pre-training decisions.
• Your approach to building dev tooling that improves research team productivity, with examples from past work.

Common mistakes to avoid

• Submitting a generic ML resume that does not mention LLM pre-training, architecture, optimizers, or large-scale training infrastructure.
• Ignoring the safety and alignment context of the company and treating the role as a standard ML engineering job.
• Claiming independent research leadership without concrete examples of projects you designed, ran, and analyzed yourself.