Periodic Labs

<h2>About Periodic Labs</h2>We're an AI and physical sciences company building state-of-the-art models to accelerate breakthroughs across materials, energy, and beyond. Backed by world-class investors and growing rapidly, we operate at the pace the frontier requires. Our team brings deep expertise, genuine ownership, and a drive to push the boundaries of what's scientifically possible.<h2>About the Role</h2>We're training frontier models to develop deep scientific knowledge and reasoning for scientific tasks. You’ll study how RL scales with training compute, develop better algorithms, and take ideas from controlled experiments to our largest runs like <a target="_blank" rel="noopener noreferrer ugc" href="https://periodic.com/news/nature-is-our-learning-environment">Periodic Neon</a>.<h2>What You'll Do</h2><ul><li>Design experiments to understand how RL performance scales with compute, model size, data, and reward quality, building on work such as<a target="_blank" rel="noopener noreferrer ugc" href="https://arxiv.org/abs/2510.13786"> ScaleRL</a></li><li>Develop better RL algorithms, spanning policy optimization, advantage estimation, exploration, and credit assignment for long-horizon RL tasks</li><li>Build adaptive sampling and curriculum methods that adjust task difficulty, problem selection, and the number of rollouts as models improve</li><li>Study bias and stability during RL training, including importance-sampling corrections and methods to tackle policy staleness and training–inference mismatch, as <a target="_blank" rel="noopener noreferrer ugc" href="https://www.youtube.com/watch?v=GH4JCdAAUYg">discussed here</a>.</li><li>Improve compute efficiency across training and inference through experiments with hyperparameters, such as length penalties, rollout counts, batch sizes, and update schedules.</li></ul><h2>You Will Thrive in This Role If You Have</h2><ul><li>Hands-on experience training LLMs with reinforcement learning</li><li>Strong attention to detail and rigorous approach to answer questions scientifically.</li><li>Coming up with small-scale RL setups that transfers to large-scale training runs.</li><li>Comfort working across a complex training stack to implement, debug, and test new research ideas.</li><li>We are open to offering a 1 year post-doc for this profile. Early career (final year PhD students) are encouraged to apply. </li></ul>Mechanics Minimum education: Bachelor’s degree or similar experienceLocation: Menlo Park, CA or Montreal, CanadaCompensation: $225,000-$350,000 base + equityVisa sponsorship: Yes, we sponsor visas and will do everything we can to assist in this process.<div> </div>

Research Scientist, Scaling RL

We're an AI and physical sciences company building state-of-the-art models to accelerate breakthroughs across materials, energy, and beyond. Backed by world-class investors and growing rapidly, we operate at the pace the frontier requires. Our team brings deep expertise, genuine ownership, and a drive to push the boundaries of what's scientifically possible.<div> </div><h2>About the Role</h2>We're training frontier models to develop deep scientific knowledge and reasoning for scientific discovery. As a Midtraining Research Engineer, you'll take base models and improve their scientific reasoning: curating and generating data, building evals, and running large-scale training experiments. Your work will also lay the groundwork for our pre-training efforts down the line.<div> </div><h2>What You'll Do</h2><ul><li>Identify, process, and curate novel sources of scientific data for large-scale model training.</li><li>Generate high-quality synthetic data to fill gaps in scientific knowledge and reasoning.</li><li>Build evaluations that correlate with downstream scientific task performance, working closely with RL researchers, physicists, and chemists.</li><li>Develop and apply techniques such as self-distillation and on-policy distillation to improve model capability.</li><li>Design and run large-scale training experiments, partnering with supercompute engineers to scale efficiently across thousands of GPUs.</li><li>Build tools for yourself and the team to investigate how data choices shape model intelligence.<div> </div></li></ul><h2>You Will Thrive in This Role If You Have</h2><ul><li>Experience training LLMs on curated mixes of trillions of tokens.</li><li>Experience on a dedicated evals team supporting a large production training run.</li><li>Hands-on use of self-distillation, on-policy distillation, or similar methods in a real training pipeline.</li><li>Experience with scaling laws and compute-optimal hyperparameters.</li><li>Comfort working across data, evals, and training infrastructure.<div> </div></li></ul><h2>Especially Strong Candidates May Also Have</h2><ul><li>Experience optimizing throughput and reliability for large-scale distributed training runs.</li><li>A background in AI for science or training on specialized domain data (e.g., protein, materials, or other scientific datasets).</li><li>Experience creating evals or synthetic data for non verifiable tasks and tracking performance over live runs.<div> </div></li></ul><h2>Mechanics</h2><ul><li>Minimum education: Bachelor's degree or similar experience</li><li>Location: Menlo Park, CA</li><li>Compensation: $250,000–$350,000 + equity</li><li>Visa sponsorship: Yes, we sponsor visas and will do everything we can to assist in this process.</li></ul>

Research Scientist / Research Engineer, Midtraining

About Periodic LabsWe're an AI and physical sciences company building state-of-the-art models to accelerate breakthroughs across materials, energy, and beyond. Backed by world-class investors and growing rapidly, we operate at the pace the frontier requires. Our team brings deep expertise, genuine ownership, and an insatiable drive to push the boundaries of what's scientifically possible.About the RoleAt Periodic Labs, we are automating scientific research in materials discovery; the data we collect, and how we represent it, is what makes this possible. We are hiring a materials scientist to work across the entire process: to find where our agents and data fail, and to fix those failures at the source. This is a hybrid research and infrastructure role. Roughly half your time will be spent working directly with our experimental and computational scientists, building agents and tools that solve active research problems. The other half will be spent improving the data those agents depend on.This role requires domain depth, not just data engineering skill. You need to know, for example, what metadata actually matters for a given characterization technique. The research half of the role keeps you grounded in the problems we are actually using LLMs to solve, so the schemas you build serve real research rather than an abstraction of it. The work demands attention to detail and a willingness to get into the weeds: to meticulously read, understand, and improve our data. You'll be as much a materials scientist doing research as the person who makes our agentic harness actually work.What You'll Do<ul><li>Work directly with lab scientists and the computational team on active research problems, staying close to the actual bottlenecks LLM agents are meant to solve.</li><li>Investigate agent traces to identify data errors and agentic failure modes, and eliminate them at the source.</li><li>Restructure and re-architect our materials databases (lab experiments, characterization data, computations) around how LLM agents actually reason and fail, not just around human readability. Draw on domain expertise about how to represent the data and what metadata matters.</li><li>Work with the hardware and automation teams to make data collection more robust.</li><li>Build research software for lab environments, translating scientific requirements into working tools.</li><li>Distill what you learn into evaluations that measure agent performance.</li></ul>Mechanics<ul><li>Minimum experience: 4+ years of research or industry experience in experimental or computational materials science and working with data at scale.</li><li>Minimum education: PhD in Materials Science, Chemistry, or a related field, or equivalent industry experience.</li><li>Location: Menlo Park, CA</li><li>Compensation: $250,000-350,000 + equity</li><li>Visa sponsorship: Yes, we sponsor visas.</li></ul>

Periodic Labs

Open jobs at Periodic Labs

Research Scientist, Scaling RL

Research Scientist / Research Engineer, Midtraining

Research Scientist / Research Engineer, Materials Data

Working for Periodic Labs?

Featured companies