Research Engineer / Research Scientist, RL Frontiers

We need to output HTML fragment with sections for headings: About the role, Responsibilities, Requirements, Qualifications, Skills, Benefits, Pay, Schedule. Only include sections that have real specific content. We must not repeat job title. Must not include Job ID etc. Must preserve all substantive info, no invention, no summarizing away detail. Remove boilerplate, navigation, legal, EEO, scam warning, duplicated lines, application instructions. We need to parse the posting. Key parts: - About Anthropic (mission etc) – maybe not needed as heading? The instruction: Use

for section headings (e.g. About the role, Responsibilities, Requirements, Qualifications, Skills, Benefits, Pay, Schedule). So we should include sections only if posting has real specific content for them. The posting includes "About The Role" (that's a heading). Also "Key responsibilities". Also "Minimum Qualifications". Also "Preferred Qualifications". Also "Representative projects". Also "The annual compensation range for this role is listed below." Also "Logistics" includes Minimum education, Required field of study, Minimum years of experience, Location-based hybrid policy, Visa sponsorship, etc. Also "We encourage you to apply..." etc (but that's boilerplate). Also "How We're Different" etc (maybe not needed). Also "Come work with us! Anthropic is a public benefit corporation..." includes benefits description. We need to map to sections: - About the role: content from "About The Role" paragraph(s). That includes description of RL Scaling team, role sits across research and engineering, etc. - Responsibilities: from "Key responsibilities" bullet list. - Requirements: maybe combine Minimum Qualifications? Or "Minimum Qualifications" could be Requirements. Also "Preferred Qualifications" could be Qualifications? Actually they have both Minimum and Preferred Qualifications. The instruction list includes Requirements, Qualifications, Skills. We need to decide mapping. Likely: - Requirements: Minimum Qualifications (must-have) - Qualifications: Preferred Qualifications (nice-to-have) - Skills: maybe from Representative projects? Or maybe skills are embedded in qualifications. Could also extract skills like "Strong programming skills in Python and JAX or PyTorch", etc. But those are already in Minimum Qualifications. Might be redundant. We need to avoid thin sections. Let's see what content we have: About The Role: multiple sentences. Key responsibilities: bullet list (6 items). Minimum Qualifications: bullet list (6 items). Preferred Qualifications: bullet list (8 items). Representative projects: bullet list (5 items) describing example projects. Compensation: Annual Salary $500,000—$850,000 USD. Logistics: Minimum education, Required field of study, Minimum years of experience, Location-based hybrid policy, Visa sponsorship. Benefits: later paragraph: "We offer competitive compensation and benefits, optional equity donation matching, generous vacation and parental leave, flexible working hours, and a lovely office space in which to collaborate with colleagues." Schedule: maybe from Location-based hybrid policy: "Currently, we expect all staff to be in one of our offices at least 25% of the time." That's schedule info. Thus we can have sections: - About the role (h3) - Responsibilities (h3) (from Key responsibilities) - Requirements (h3) (from Minimum Qualifications) - Qualifications (h3) (from Preferred Qualifications) - Skills (h3) maybe from Representative projects? But Representative projects are more like example tasks, not skills. Could also include skills from qualifications but that duplicates. Perhaps we can omit Skills if no distinct content. But we need to follow instruction: include a section ONLY when posting has real, specific content for it. If we don't have distinct skills beyond what's in qualifications, we could omit Skills. - Benefits (h3) from benefits paragraph. - Pay (h3) from compensation. - Schedule (h3) from hybrid policy. We must not include headings with filler like "commensurate with experience". We have actual numbers for pay, schedule info. We must preserve ALL substantive information; do not invent or summarize away detail. So we need to keep all details from each section, not summarizing. Use

• for any list of duties/requirements/benefits, and

  for prose. Thus for each section we need to output either

  prose or

  • list. We must not repeat job title. Not include Job ID etc. We must remove boilerplate: navigation, legal, EEO, scam warning, duplicated lines, application instructions. So we need to cut out paragraphs like "We encourage you to apply even if you do not believe you meet every single qualification..." etc. Also "How We're Different" etc. Also "Come work with us! Anthropic is a public benefit corporation..." partially we need benefits but we can extract benefits sentence; the rest maybe boilerplate. The sentence about benefits is substantive: "We offer competitive compensation and benefits, optional equity donation matching, generous vacation and parental leave, flexible working hours, and a lovely office space in which to collaborate with colleagues." That's benefits. The rest of that paragraph maybe fluff but we can keep only that sentence? The instruction: Preserve ALL substantive information; do NOT invent or summarize away detail. So we need to keep the substantive info; we can discard fluff that is not substantive? The sentence about benefits is substantive. The rest like "Anthropic is a public benefit corporation headquartered in San Francisco." maybe not needed for benefits section but could be considered company info; but not required headings. We can omit as not part of any section. Similarly "Guidance on Candidates' AI Usage: Learn about our policy for using AI in our application process." is boilerplate. Thus we will produce HTML with sections:

    About the role

    ... (the About The Role paragraph(s) )

    Responsibilities

    • Study how RL training and sampling scale with model size, context length, and compute, and find the algorithmic and systems changes that keep scaling efficient
    • Develop next-generation model architectures and RL algorithms, and make them run efficiently at frontier scale
    • Take promising small-scale results to frontier-scale runs, and diagnose why they behave differently when they get there, whether the cause is numerical, algorithmic, or systemic
    • Build the experimental infrastructure that sets research velocity: fast, reproducible comparisons of architecture and algorithm variants at meaningful scale
    • Own end-to-end performance of our largest RL runs, from research code down to the hardware
    • Build performance and cost models for proposed architecture and algorithm changes, and use them to decide which ideas get scaled
    • Investigate training dynamics at scale, including instabilities, divergence, and throughput regressions, and trace them to root cause
    Note: The bullet list originally had 7 items? Let's count: The text shows: Key responsibilities Study how RL training and sampling scale with model size, context length, and compute, and find the algorithmic and systems changes that keep scaling efficientDevelop next-generation model architectures and RL algorithms, and make them run efficiently at frontier scaleTake promising small-scale results to frontier-scale runs, and diagnose why they behave differently when they get there, whether the cause is numerical, algorithmic, or systemicBuild the experimental infrastructure that sets research velocity: fast, reproducible comparisons of architecture and algorithm variants at meaningful scaleOwn end-to-end performance of our largest RL runs, from research code down to the hardwareBuild performance and cost models for proposed architecture and algorithm changes, and use them to decide which ideas get scaledInvestigate training dynamics at scale, including instabilities, divergence, and throughput regressions, and trace them to root cause That's 7 items (I see 7 sentences). We'll list each as
  • .

    Requirements

    • Deep familiarity with modern transformer language models, including their architecture, training dynamics, and the behavior of large-scale optimization
    • Hands-on experience training large models in a distributed setting, including the tradeoffs between data, tensor, and pipeline parallelism
    • A track record of original technical work in ML training or systems, such as new methods, architectures, or optimizations, demonstrated through research, open-source, or production impact
    • Ability to design rigorous experiments at scale, including baselines, ablations, and enough statistical care to trust a result that costs real compute
    • Ability to reason quantitatively about the compute, memory, and communication costs of a model or algorithm
    • Strong programming skills in Python and JAX or PyTorch, and comfort reading and changing code at every layer of the stack

    Qualifications

    • Research experience in reinforcement learning, optimization, or large-scale training, published or otherwise
    • Experience developing RL algorithms for language models
    • Experience with scaling laws or other quantitative models of training efficiency
    • Experience designing or modifying transformer architectures beyond standard configurations
    • Experience scaling training to large fleets of accelerators and debugging the problems that only appear at scale
    • Deep understanding of numerics in large-scale training, including low-precision formats and sources of instability
    • Familiarity with how GPU or TPU performance characteristics shape architecture and algorithm choices
    • Experience with C++ or Rust
    We could also include Representative projects as a section maybe "Projects" but not in headings list. Could be considered under Skills? But we don't have Skills heading. We could add a section "Representative projects" but not in allowed headings list; instruction says use

    for section headings (e.g. About the role, Responsibilities, Requirements, Qualifications, Skills, Benefits, Pay, Schedule). It says e.g., not exclusive list, but we should prefer those headings. However we could add a section for Representative projects if we think it's substantive and not covered elsewhere. But maybe we can incorporate them under Skills or Qualifications? They are example projects, not skills. Could be considered under "Skills" as demonstration of ability. But better to follow instruction: prefer a few substantive sections over many thin ones. We already have many sections; adding another may be thin. However Representative projects are substantive detail about what you'll do. Could be considered under Responsibilities? Not exactly. Might be better to include as a separate section "Projects" but not in the list; risk of deviating. The instruction: "Use

    for section headings (e.g. About the role, Responsibilities, Requirements, Qualifications, Skills, Benefits, Pay, Schedule)." The e.g. suggests those are examples, not exhaustive. So we can add other headings if needed. But they said "Prefer a few substantive sections over many thin ones." Adding a Projects section may be okay. But we already have Responsibilities covering what you'll do; Representative projects are examples of the type of work. Could be considered part of Responsibilities. Might be okay to omit to keep sections fewer. We'll skip Representative projects. Now Benefits:

    Benefits

    We offer competitive compensation and benefits, optional equity donation matching, generous vacation and parental leave, flexible working hours, and a lovely office space in which to collaborate with colleagues.

    Pay:

    Pay

    Annual Salary $500,000—$850,000 USD

    Schedule: We have location-based hybrid policy: "Currently, we expect all staff to be in one of our offices at least 25% of the time. However, some roles may require more time in our offices." Thus schedule info: expectation of office presence at least 25% of time. Thus:

    Schedule

    Currently, we expect all staff to be in one of our offices at least 25% of the time; some roles may require more time in our offices.

    We should also include Minimum education etc under Logistics; but we have no heading for