Scientific Software Engineer - AI/ML for Hyperspectral Imaging

Details

Posted: 2026-05-28

Location: Berkeley, California

Lawrence Berkeley National Laboratory's (Berkeley Lab) Advanced Light Source (ALS) Division has an opening for a Scientific Software Engineer specializing in AI/ML for hyperspectral imaging. This role advances AI-driven scientific discovery by developing machine learning methods and scalable data analysis tools for complex, high-dimensional scientific datasets.

The engineer will build and generalize segmentation, feature extraction, and modeling workflows, including development of a foundation model to extract scientific information from hyperspectral imaging data across infrared imaging, resonant soft X-ray scattering, tomography, and ptychography.

Key responsibilities:

• Expand and generalize AI-driven segmentation and feature extraction workflows across multiple scientific modalities and domains.
• With general guidance, develop and apply specialized machine learning models for hyperspectral imaging data, serving as a key target domain for high-dimensional spectral-spatial analysis.
• Operating under broad direction, develop interfaces and data products that enable machine learning models to be integrated into higher-level automation and agent-based systems.
• Implement scalable pipelines that transform experimental data into structured, semantically meaningful scientific representations.
• Ensure reproducibility, traceability, and interoperability of software and AI workflows across systems and facilities.
• Collaborate with scientists and engineers to gather requirements, validate results, and translate scientific needs into software solutions.
• Design, test, deploy, and maintain robust software using modern development practices (e.g., CI/CD, version control, unit testing).
• Contribute to open-source projects, develop documentation, provide user support, and communicate work through presentations.

Required qualifications:

• Bachelor's degree and a minimum of 2 years of related experience; or an advanced degree without experience (Master's or PhD); or equivalent years of work experience.
• Experience with the open-source scientific Python ecosystem (e.g., NumPy, PyTorch, TensorFlow, scikit-learn).
• Hands-on experience analyzing complex scientific datasets, including imaging, multivariate, multimodal, multichannel, or volumetric data.
• Hands-on experience developing, training, or applying AI/ML models, including segmentation methods, for scientific data analysis.
• Experience developing or contributing to software projects, including collaborative or open-source development.
• Experience building or maintaining data analysis pipelines or scientific workflows.
• Ability to work collaboratively with a team of scientists and engineers.
• Knowledge of AI/ML principles and data analysis methods relevant to complex scientific data, including segmentation, feature extraction, model training, validation, and interpretation.
• Knowledge of GPU acceleration and performance profiling for large scale workflows
• Demonstrated ability to design and evaluate workflows for processing, analyzing, and representing complex scientific imaging and high-dimensional data.
• Proficiency to validate data quality, model outputs, and workflow results against technical and scientific expectations.
• Proven capability to develop, test, debug, document, and maintain reproducible software and machine learning workflows.
• Effectiveness in communicating technical results clearly, both in writing and verbally, to interdisciplinary audiences.
• Flexibility and capacity to learn new scientific domains, data modalities, tools, and computational techniques within evolving project timelines.

Desired skills/knowledge:

• Experience with hyperspectral scientific datasets.
• Experience with High-Performance Computing (HPC) environments.
• Experience with MLOps tools such as MLflow.
• Experience with CI/CD tools (e.g., GitHub Actions).
• Familiarity with hyperspectral imaging data.
• Familiarity with agent-based or AI orchestration frameworks (e.g., LLM-based or multi-agent systems).