DARPA researchers modeling complex biological systems now have access to a dedicated AI and high-performance computing environment to accelerate large-scale simulations and analysis.
Parallel Works and CoreWeave have deployed the managed AI platform for the agency’s Network of Optimal Dynamic Energy Signatures (NODES) program.
The setup gives NODES teams rapid access to the computing resources needed to run biological simulations and train AI models. Researchers can begin computational work within days rather than wait weeks or months for infrastructure to be deployed and configured.
Parallel Works operates the environment through its ACTIVATE platform, which handles researcher onboarding, identity and access, resource provisioning, scheduling, and operational reporting.
The company also provides direct technical support for both the platform and research applications.
The environment includes dedicated NVIDIA HGX H100 systems, enterprise-scale storage, and high-speed NVIDIA Quantum InfiniBand networking.
ACTIVATE divides the reserved computing capacity among NODES research teams, giving each a guaranteed allocation while allowing them to draw from shared resources when additional capacity is available.
One Platform, Multiple Workloads
Through a single platform, researchers can also access CoreWeave Kubernetes Service and SUNK, its Slurm-on-Kubernetes offering. This allows teams to run workloads of different types and sizes within the same computing environment.
ACTIVATE tracks resource usage and utilization across the program, while CoreWeave monitors the underlying infrastructure to identify issues before they interrupt long-running research workloads.
“Breakthrough science shouldn’t be delayed by infrastructure,” Parallel Works chief executive Matthew Shaxted said.
“DARPA needs researchers focused on biology, not on building infrastructure or managing compute environments.”
