AI news ·
Berkeley Lab architect calls for scientists to take on engineering roles to build AI-ready data platforms
Government labs must build AI-ready data infrastructure before deploying machine learning on scientific problems, a Berkeley Lab architect says. A workforce gap between scientists and engineers is slowing adoption.
Berkeley Lab scientist outlines data infrastructure needs for AI research
Dr. Patrick Huck, Principal Platform Architect at Lawrence Berkeley National Laboratory, said government organizations need to build data foundations designed for AI before deploying machine learning tools on scientific problems.
Huck pointed to the Materials Project as a working example. The platform serves 730,000 registered users with 5,000 daily active researchers. It integrates materials science data across multiple domains and uses AI to accelerate discovery.
The gap between what scientists can do and what engineers can build is widening, Huck said. Scientists need engineering and architectural skills to design platforms and manage data at scale.
Workforce mismatch slows adoption
Government labs face a shortage of people who can bridge scientific research and engineering work. Huck recommended creating dedicated data reliability engineering teams to fill this gap and support principal investigators.
Performance evaluations should reward scientists who enable others' research, not just individual discoveries. This shift would encourage the platform-building work that makes AI-ready data possible.
Data curation is the bottleneck
Building AI systems requires clean, well-organized data. Scientists often lack the skills or time to structure their own datasets for machine learning. Organizations that invest in data engineering see faster AI adoption across research teams.
For researchers looking to develop these capabilities, AI Data Analysis Courses and AI Research Courses cover the technical foundations needed to work with large scientific datasets.