AI and Data Assimilation Scientist

Overview

Lynker is seeking a talented and experienced AI and Data Assimilation Scientist to support the Environmental Modeling Center (EMC) within the National Centers for Environmental Prediction (NCEP). The primary objective of this role is to develop and transition innovative AI-based Data Assimilation (DA) systems. These systems will complement existing physics-based systems and be tested as independent prototypes, running alongside traditional DA workflows.

The position is located at the NOAA Center for Weather and Climate Prediction (NCWCP) in College Park, MD.

Duties of the AI and Data Assimilation Scientist will include the following:

The AI and Data Assimilation Scientist will perform their job duties to a high standard, working both independently and collaboratively, focusing on scientific developments that advance the use of AI-DA techniques as an alternative to traditional ensemble/variational-based techniques. The core responsibility is to lead the development, implementation, testing, and evaluation of an AI-based Real-Time Mesoscale Analysis (AI-RTMA) system in support of NOAA’s National Blend of Models (NBM).

The AI-RTMA system will generate high spatial and temporal resolution analyses of meteorological variables to reduce biases in the NBM fields. Because these fields serve as the foundation for gridded forecasts issued by the National Weather Service, this system will directly contribute to improved forecast quality.

The successful AI and Data Assimilation Scientist will work on the following scientific and engineering tasks:

• Conduct a comprehensive review of state-of-the-art AI-based data assimilation and end-to-end weather forecasting methodologies, systems, and frameworks. Communicate findings with EMC scientists and external partners to inform the development of a scientifically robust and efficient AI-RTMA approach.
• Collaborate with NOAA’s NBM team and key stakeholders to define product requirements for AI-RTMA, including domain configuration, grid structure, output variables, spatial and temporal resolution, and data formats suitable for operational evaluation and transition.
• Design, implement, and maintain robust data pipelines to support AI-RTMA training, validation, testing, and evaluation. This includes collecting, formatting, quality-controlling, and integrating diverse observational datasets (e.g., conventional observations, satellite, radar, and other sources), as well as preparing model inputs, targets, metadata, and training/validation splits.
• Develop, train, rigorously test, and deploy a fully functional AI-RTMA system based on selected AI frameworks or architectures.
• Implement cross-validation and other evaluation methodologies to quantify model performance and reliability during inference.

The AI and Data Assimilation Scientist selected should have the following:

• Experience developing, training and deploying AI-based systems applied to geophysical systems.
• Experience with common AI frameworks such as PyTorch, Tensor Flow.
• Experience working with earth observation data, including conventional observations, satellite, radar.
• In-depth knowledge of data assimilation techniques (observation forward modeling, quality control, variational-based and/or ensemble methods).
• Strong foundation in the physical, statistical and mathematical basis of geophysical modeling (atmospheric and/or environmental).
• Excellent Python programming skills.
• Practical experience utilizing High Performance Computers (HPCs) and GPUs.
• Proven experience working in a UNIX environment with advanced scripting languages.
• Good communication skills, both oral and written, in English.

The Ideal AI and Data Assimilation Scientist will have the following:

• Experience with cloud platforms and use of IDEs for development.
• Experience with cloud-native data formats such as Zarr, Parquet.
• Experience with compiled languages.
• Comfort using agentic AI tools to accelerate development.
• Experience executing numerical models on HPC platforms using parallelization frameworks and job scheduling systems.
• Familiarity with coupled earth system models.
• Knowledge of modern software engineering practices…