The NERDD microservices platform for deploying machine learning models in drug discovery

Machine learning models are transforming data-driven research across scientific disciplines, yet their deployment as accessible and reliable web services remains a significant challenge. We introduce the NERDD microservices platform, a scalable, maintainable, and secure framework designed to support the sustainable delivery of machine learning services in cheminformatics, bioinformatics, and related domains. Built in Python, NERDD provides a modular infrastructure for deploying predictive and generative models, with native support for chemistry-specific tasks such as molecular property prediction, atom-level inference, and molecular generation. Through standardized configuration files and containerized modules, the platform enables seamless integration of in silico models and interaction via a REST API and an intuitive web interface. Leveraging technologies like Apache Kafka, Kubernetes, and Ceph, NERDD ensures high availability, fault tolerance, and dynamic scalability. As an open-source solution, it empowers researchers in drug discovery to deploy machine learning models in a reproducible and maintainable manner. The NERDD microservices platform is an open‑source, chemistry‑aware framework for deploying predictive and generative models as web services and APIs, coupling a modern backend with Kafka‑based messaging, Kubernetes orchestration, and KEDA‑driven autoscaling to ensure scalable, highly available operation. NERDD streamlines integration and execution, providing configuration‑driven pipelines with automatic input handling and a processing workflow with checkpoints and idempotent handling for robust, reproducible results.