Member of Technical Staff – Post-Training, Reinforcement Learning, Generalization

This is a fully remote position, open to applicants in California, +2 more states.

📋 Description

• Train large language models utilizing reinforcement learning techniques.

• Create methodologies that enhance generalization beyond the training distribution.

• Develop training recipes, environments, and evaluation metrics to assess acquired skills.

• Conduct and analyze large-scale reinforcement learning post-training experiments, including RLHF and RL from verifiable rewards.

• Investigate reward design, as well as environment and task diversity, curricula, and regularization.

• Construct held-out evaluations for previously unseen tasks, domains, and formats.

• Identify reward hacking and benchmark overfitting issues.

• Diagnose training dynamics and failure modes, including instability, reward over-optimization, and capability regressions.

• Transform successful research into robust, reproducible production-model training protocols.

• Collaborate with teams focused on post-training data infrastructure, evaluation, and pre-training.

⛳️ Requirements

• Strong research and software engineering abilities, with practical experience in training large language models.

• Experience in applying reinforcement learning techniques to LLMs, such as PPO, GRPO, RLHF, or RLVR, preferably at scale.

• Proven track record in generalization, robustness, or distribution shift through publications or deployed models.

• Methodical experimental practices, including clean ablations, careful controls, and sound conclusions drawn from noisy results.

• Familiarity with distributed training frameworks like JAX, PyTorch, FSDP, or Megatron is advantageous.

• Excellent verbal and written communication skills.

• Profound understanding of data quality principles and best practices.

• Readiness to delve deeply into data to evaluate dataset quality and its effects on model training outcomes.

🏝️ Benefits

• Equity offerings available.

• Weekly lunch stipend of $75/£75 or equivalent in local currency.

• Comprehensive health and dental benefits.

• Dedicated budget for mental health support.

• RRSP matching, 401K, and Pension Scheme options.

• 100% parental leave top-up for up to 6 months for either parent.

• Annual benefits for arts and culture, fitness/wellness, quality time, and workspace improvement.

• Education and learning stipend for conferences, courses, and coaching opportunities.

• 6 weeks of paid vacation (30 working days).

• Budget for travel to other offices if working remotely.

• Annual company offsite events.

• Daily lunch program, snacks, and regular community and social events for office-based employees.

• Co-working benefits for employees who are not located near an office.

• $500 stipend for home office setup.