Applied Research – Scientific AI

This is a fully remote position, open to applicants in United States, +2 more locations.

📋 Description

• Develop and fine-tune agents for scientific research, which includes literature searches, executing code, utilizing domain tools, and selecting experiments.

• Collaborate across autonomous research and computational science with environments and evaluations tailored to research partners' tools, data, and experimental insights.

• Engage directly with customers, users, or research partners while upholding a significant research agenda.

• Create long-term environments for literature searches, scientific database inquiries, code execution, domain tool utilization, and experiment selection within predetermined budgets.

• Transform research partners' workflows, datasets, and outcome criteria into trainable tasks, evaluations, and reward signals.

• Design verifiers to ensure correctness, novelty, and reproducibility, including checks for label leakage, shortcut lookups, and unsupported conclusions.

• Integrate programmatic checks, simulations, domain models, and expert feedback into training; adjust confidence estimates and establish human-review requirements.

• Develop RL and multi-agent methodologies for proposing, testing, and critiquing hypotheses.

• Benchmark methods against robust single-agent baselines using equivalent computational resources and assess performance on held-out problems.

• Ensure reproducibility of code, data, logs, and experiment configurations; re-execute results, measure noise, conduct ablations, and release reusable environments, evaluations, and research code.

⛳️ Requirements

• Expertise in RL, agents, or post-training systems, with the capability to grasp relevant scientific contexts.

• Previous experience in autonomous research, ML for science, scientific computing, or reproducible computational research is advantageous but not mandatory.

• Practical experience in building and troubleshooting model-training, post-training, agent, evaluation, or closely related experimental processes.

• Strong experimental judgment: meaningful baseline establishment, meticulous trace inspection, clear failure attribution, and readiness to shift direction based on evidence.

• Proficiency in Python.

• Ability to create artifacts that other researchers and engineers can execute, inspect, and build upon.

• Comfort in managing ambiguous challenges while keeping partners and teammates informed.

• Demonstrated excellence through publications, open-source projects, benchmarks, or delivered products.

🏝️ Benefits

• Equity incentives.

• Flexible work options (San Francisco, New York, London, or remote).

• Visa sponsorship and relocation assistance.

• Professional development budget.

• Team off-sites and conference participation.