NewsBreak
Research Intern, Agent RL Training
Mountain View, California, United States · Internship · Posted May 26, 2026
About the job
Conduct research and experiments applying large language models to NewsBreak products, including content understanding, recommendations, web browsing, and multi-step tasks.
What you'll do
- Identify research directions with a mentor and run end-to-end supervised fine-tuning experiments on LLM-based agents; assist with RL exploration.
- Build training datasets, including instruction-following examples, preference pairs, agent trajectories, and synthetic data.
- Contribute to research publications and potential top-venue submissions.
What they ask for
- Independently conduct end-to-end model SFT and have a basic understanding of RL-based post-training methods such as RLHF, DPO, PPO, and GRPO.
- Strong Python and PyTorch skills.
- Ability to assess and explain desirable model behavior in user-facing domains.
- Motivation and commitment to research, including willingness to put in extra hours when needed.
Nice to have
- Publication at a top-tier venue such as NeurIPS, ICML, ICLR, ACL, or EMNLP.
- Experience with distributed training, custom GPU kernels, synthetic-data pipelines, or open-source RL frameworks.
Pay
$35- $50 USD hourly pay; $35 — $50 USD annual base pay range
Tools
LLMs, Python, PyTorch, RLHF, DPO, PPO, GRPO, Triton
AI focus
Research and train LLM-based agents, including supervised fine-tuning, RL-related exploration, reward design, and agent capabilities.
Model-written summary · the original posting is authoritative