Research Scientist

The Nuclear Company•Washington, DC

4d•$150,000 - $173,000

About The Position

The Nuclear Company is seeking a Research Scientist to join its Applied Research and AI team. This role involves tackling complex research problems at the intersection of AI and large-scale infrastructure, focusing on optimizing construction, capital allocation under uncertainty, and securing distributed critical infrastructure. The position offers the opportunity to research novel approaches, collaborate with domain experts and engineering partners, and deploy systems that inform operational decisions across construction, capital planning, and security. The team consists of top nuclear industry experts, PhD-level researchers, and software engineers dedicated to productionizing models.

Requirements

PhD in Computer Science, Machine Learning, Operations Research, Economics, Applied Mathematics, or a closely related quantitative field; or MS with a demonstrable track record of independent research output (publications, patents, or equivalent deployed systems).
Hands-on experience implementing and evaluating deep RL algorithms; fluency in policy gradient methods (PPO, TRPO, SAC), value-based approaches (DQN variants, IQL), and the tradeoffs between model-free and model-based RL.
Experience building RL training environments; demonstrated ability to translate complex real-world operational processes into tractable MDP formulations with appropriate state/action/reward design.
Production-quality Python; deep learning frameworks (PyTorch); version control, testing, and reproducibility practices.
Demonstrated ability to operate in a fast-moving environment where problem definitions evolve, priorities shift, and hands-on technical contribution is expected.

Nice To Haves

Offline / Batch RL experience (IQL, CQL, TD3+BC, Decision Transformer, or similar methods).
Combinatorial Optimization + ML experience (Graph neural networks for scheduling or routing, neural combinatorial optimization, or hybrid learned/exact solver approaches).
Multi-Agent RL experience (MADDPG, QMIX, MAPPO, or related methods).
Stochastic / Robust Optimization experience (CVaR-constrained RL, distributionally robust MDPs, or chance-constrained programming).
Experience monitoring and retraining RL systems post-deployment, including distribution shift detection and safe policy update procedures.
Domain exposure in construction project management, infrastructure operations, energy industry, electricity markets, nuclear power, industrial control systems, or physical/cyber security for critical infrastructure.
Familiarity with game-theoretic frameworks (strategic interactions, mechanism design, adversarial dynamics, or equilibrium concepts).
Understanding of behavioral science (cognitive biases, bounded rationality, responses to uncertainty).
2-5 years of post-PhD research experience with demonstrated ownership of a research program.
Publications at top-tier venues (NeurIPS, ICML, ICLR, AAMAS, or equivalent).
Experience driving a team’s external research agenda and identifying contributions for external dissemination.
Experience formally mentoring PhD-level researchers or interns.
Experience making research architecture decisions that span research and engineering.
Experience communicating technical direction to non-technical senior stakeholders.

Responsibilities

Translate complex operational processes into well-defined research problems and identify appropriate modeling approaches.
Build simulation environments to represent operational processes (construction scheduling, portfolio sequencing, security operations) for training and evaluating decision-making models.
Design rigorous experiments, maintain reproducible codebases, and communicate results in internal reports and external publications.
Develop models to optimize construction scheduling across multiple concurrent sites, minimizing schedule variance, resource idle time, and cascading delays.
Design adaptive scheduling approaches that respond to real-time disruptions and learn from historical project data.
Build models to inform site development sequencing and capital allocation across a growing fleet, considering regulatory milestones, capital constraints, and correlated risks.
Develop approaches to account for uncertainty in key inputs (permitting timelines, cost distributions, grid demand forecasts) for portfolio decisions with bounded downside.
Build models for alert prioritization, anomaly detection, and patrol scheduling to support physical and cyber security operations.
Design systems where models and human analysts share decision authority, communicating uncertainty clearly and degrading safely.
Collaborate with engineering to define how models are served, monitored, updated, and overridden in production.
Present research results and system performance to stakeholders and translate findings into actionable operational recommendations.