Applied AI/ML Engineer (Founding Team)

Riven Systems; New York (NY); On-site; USD 120,000 - 180,000; Full time

Job summary

Riven Systems in New York is seeking a Machine Learning Scientist to help define multi-scale modeling futures, working with thermodynamics, hydrometallurgy, and autonomous experimentation to merge ML with physical modeling.

You will build learned components of thermodynamic and process models, design active learning experiments, and develop optimization methods for flowsheet design at scale, with a strong emphasis on uncertainty quantification.

Qualifications

• Ph.D. in machine learning, applied mathematics, statistics, physics, chemical engineering, or related quantitative field or equivalent industry experience.
• Substantial experience with Bayesian methods, Gaussian processes, or comparable approaches to modeling under uncertainty with limited data.
• Practical command of optimization, from gradient-based training through constrained nonlinear programming.
• Experience building with LLM agents, including tool use, evaluation design, and the failure modes of agentic systems.
• Comfortable with the tools and processes for building large software systems.
• Clear interdisciplinary communication and the curiosity to quickly learn and execute in new domains.
• Ability to prioritize, execute, and collaborate with a high degree of agency, a willingness to be wrong, and the ambition to build something that has never existed before.
• Trust, humor, rigor, and an excitement to work in a scrappy, interdisciplinary early-stage environment.

Responsibilities

• Build the learned components of Riven’s thermodynamic and process models, and the methods for deciding where learning helps and where physical structure should hold.
• Own the optimal experimental design that drives our autonomous lab: active learning, Bayesian optimization, and information-based selection of what to measure next.
• Develop optimization methods for flowsheet design and operation, including surrogate-assisted and derivative-based approaches at scale.
• Build the agentic systems that let LLMs propose, simulate, and revise chemical process designs against our models.
• Design the evaluations that establish whether those agents are actually right.
• Quantify and propagate uncertainty across the stack, from parameters through predictions to design decisions.

Skills

Bayesian methods

Optimization

LLM agents

Software engineering

Interdisciplinary communication

Education

Ph.D. in machine learning / related field

Tools

JAX

PyTorch

Job description

About the role

As a Machine Learning Scientist you will help define the future of multi-scale modeling. You will work closely with experts in thermodynamics, hydrometallurgy, process modeling, and autonomous experimentation to decide how to unite modern ML and AI with rigorous physical modeling.

Riven is hiring a lean, interdisciplinary team and experience spanning multiple roles is welcome. Apply to whichever role is closest to your skillset.

Initial Responsibilities

• Build the learned components of Riven’s thermodynamic and process models, and the methods for deciding where learning helps and where physical structure should hold.
• Own the optimal experimental design that drives our autonomous lab: active learning, Bayesian optimization, and information-based selection of what to measure next.
• Develop optimization methods for flowsheet design and operation, including surrogate-assisted and derivative-based approaches at scale.
• Build the agentic systems that let LLMs propose, simulate, and revise chemical process designs against our models.
• Design the evaluations that establish whether those agents are actually right.
• Quantify and propagate uncertainty across the stack, from parameters through predictions to design decisions.

Qualifications

• Ph.D. in machine learning, applied mathematics, statistics, physics, chemical engineering, or a related quantitative field, or equivalent industry experience.
• Substantial experience with Bayesian methods, Gaussian processes, or comparable approaches to modeling under uncertainty with limited data.
• Practical command of optimization, from gradient-based training through constrained nonlinear programming.
• Experience building with LLM agents, including tool use, evaluation design, and the failure modes of agentic systems.
• Comfortable with the tools and processes for building large software systems.
• Clear interdisciplinary communication and the curiosity to quickly learn and execute in new domains.
• Ability to prioritize, execute, and collaborate with a high degree of agency, a willingness to be wrong, and the ambition to build something that has never existed before.
• Trust, humor, rigor, and an excitement to work in a scrappy, interdisciplinary early-stage environment.

Bonus Points

• Physics-informed machine learning, neural differential equations, or learned surrogates for scientific simulation.
• Optimal experimental design, active learning, or closed-loop autonomous experimentation.
• Automatic differentiation frameworks (JAX, PyTorch) applied to scientific rather than deep-learning workloads.
• Reinforcement learning, LLM post-training, or eval-driven development.
• Background in thermodynamics, chemistry, materials science, or process systems engineering.
• Contributions to open-source scientific machine learning tooling.