2 days ago
Base Salary
$220k - $357k/yr
Responsibilities
- Design and run rigorous experiments that produce measurable improvements in model capability.
- Improve data recipes through data quality, deduplication, and mixture experiments.
- Define high-quality RL data and rewards, including grader design and prevention of reward hacking.
- Advance training science through architecture, scaling, and post-training experiments for coding and computer-use agents.
- Build evaluations and targeted tasks that expose model weaknesses and guide research priorities.
- Partner with engineers to scale promising results from small experiments to frontier-scale training runs.
- Own research directions end to end in a small-team setting.
Requirements
- At least 3 years of hands-on experience training or post-training large neural networks with 1B+ parameters, including pre-training, fine-tuning, or RL-based methods such as RLHF or DPO.
- Production-quality programming experience in Python, PyTorch, JAX, Rust, or C++.
- Experience with RL algorithms for language models, scaling laws, or RL environments and evaluations is preferred.
- Experience with distributed training at scale and debugging issues that appear only at scale is preferred.
- Experience training frontier-scale models and strong RL and machine learning fundamentals are preferred.
- Excellent programming and software engineering ability across the research stack, especially in Python and PyTorch or JAX, is preferred.
- A track record of original ML work demonstrated through research, open source, or production impact is preferred.
- An MS or PhD in computer science, machine learning, mathematics, physics, or a related field, or equivalent research experience, is preferred.
- Must be able to work onsite at the Palo Alto office.
Benefits
- Equity is included in the total rewards package.
- Comprehensive medical, vision, and dental coverage.
- Access to a 401(k) retirement plan.
- Short- and long-term disability insurance.
- Life insurance, discounts, and other perks.
- This is an onsite role at the Palo Alto office.
Categories
About Cursor
Cursor is an AI-powered code editor and coding assistant from Anysphere, built for professional software engineers and enterprise teams. It combines a VS Code–style editor with integrated AI chat, code generation, and repo-aware refactoring, and is sold via individual, team, and enterprise subscriptions. The company also targets sectors like healthcare and state and local education with field engineers who support evaluations and proofs of concept.