SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Phaidra is building AI-powered control systems for industrial automation. The company uses reinforcement learning to enable factories, power plants, and buildings to automatically learn and improve over time, replacing hard-coded static controls with adaptive intelligent systems. The team has deep expertise in applying AI to complex real-world problems, including work from DeepMind (AlphaGo) and Google's data center cooling optimization.
As a Senior AI Research Scientist specializing in model-based reinforcement learning, you will lead research efforts to develop novel algorithmic architectures for intelligent industrial control systems. You will design, implement, and evaluate model-based RL agents including planning-based controllers (MPC, MPPI) and software prototypes for deployment on real industrial systems.
Key responsibilities include: developing learned dynamics and world models that generalize across systems with robust training pipelines (pretraining, curriculum learning, active/adversarial learning, fine-tuning); researching and implementing safe RL, constrained control, scenario planning, and Bayesian RL methods to ensure safety during deployment; reporting research findings internally and externally; organizing collaborative research projects with external partners; mentoring Research Engineers to translate research into industrial applications; independently defining new research directions; and owning development and rollout of entire research areas or large projects.
You will work with a distributed team across the USA, Canada, UK, Sweden, Spain, Portugal, the Netherlands, Singapore, Australia, and India. The company is 100% remote with no physical office and uses OysterHR for international hiring.
This role is ideal for someone with a PhD in a technical field and strong background in model-based reinforcement learning, with demonstrated expertise in planning algorithms, world models, deep learning, control theory, or safe/constrained RL.
