Applied Scientist: Predictive Modeling for Next Experiments

Nodi; New York (NY); On-site; USD 235,000 - 295,000; Full time

Benefits offered by this job

Scientific workflow orchestration

LLM agents

Job summary

Nodi is seeking a senior scientist to design and validate predictive physics, chemistry, and materials models that guide next experiments. You will collaborate with campaign scientists and ML engineers to ensure the modeling and experiment-selection logic is robust, transparent, and integrated into the agent workflow.

We value deep quantitative training, a PhD, and the ability to turn ideas into reproducible tests.

Qualifications

• Deep quantitative scientific training with hands-on model validation against experiments.
• Strong Python and scientific computing skills for building working models.
• Experience with Bayesian optimization and experimental design.

Responsibilities

• Build models that influence experimental decisions and select suitable tasks.
• Calibrate models against reference data and assess reliability.
• Automate model-improvement experiments with reproducible workflows.
• Validate improvements on held-out data and define acceptance criteria.
• Decide which experiment to run next based on predictions and costs.
• Investigate disagreements between models and laboratory data.
• Translate models into agent-ready tools with clear inputs and failure conditions.
• Learn from outcomes to guide future modeling and testing.

Skills

Python programming

Bayesian optimization

Active learning

Experimental design

Data analysis

Education

PhD in Physics, Materials Science, Chemistry, Chemical Engineering, Applied Mathematics, Scientific ML, or related field

Equivalent industrial research experience

Tools

Python

Jupyter

Job description

Nodi is seeking a senior scientist to design and validate predictive physics, chemistry, and materials models that guide next experiments. You will collaborate with campaign scientists and ML engineers to ensure the modeling and experiment-selection logic is robust, transparent, and integrated into the agent workflow.

We value deep quantitative training, a PhD, and the ability to turn ideas into reproducible tests.