ML Researcher, Apple Foundation Models

What you'll do

• Train foundation models for reasoning, coding, and agentic capabilities using reinforcement learning.
• Develop autonomous coding agents capable of operating in real repositories and handling multi-step workflows.
• Implement RL with verifiable rewards for mathematical reasoning and multi-turn RL for coding tasks.
• Research and apply scaling laws for RL compute allocation and progressive alignment across capability stages.
• Train models to manage their own context for long-horizon tasks and long context windows.
• Develop reward modeling techniques to resist reward hacking and handle sparse or delayed rewards.
• Optimize models specifically for Apple silicon to ensure private and integrated OS experiences.
• Perform on-policy and cross-stage distillation to combine independently optimized capabilities into single models.

What we're looking for

• PhD or equivalent practical experience in Computer Science or a related technical field.
• Demonstrated expertise in deep learning with publications at top ML or NLP conferences or a track record of applying deep learning to products.
• Proficient programming skills in Python and one deep learning toolkit such as JAX, PyTorch, or Tensorflow.
• Ability to work in a collaborative environment.
• Experience with reinforcement learning for LLMs, including RLHF, GRPO, PPO, RLVR, reward modeling, and RL scaling laws (preferred).
• Experience with code generation, coding agents, and repository-level code understanding (preferred).
• Experience with agentic systems, multi-turn RL, tool-use planning, and long-horizon task execution (preferred).
• Experience with distillation, alignment, long context, and efficiency techniques (preferred).