Researchers have introduced the NERDD microservices platform, a scalable, maintainable, and secure framework designed to support the sustainable delivery of machine learning services. While machine learning models are increasingly transforming data-driven research across diverse scientific disciplines, deploying these computational tools as accessible and reliable web services has remained a significant challenge. The newly developed platform provides an open-source solution structured to address these deployment obstacles directly within cheminformatics, bioinformatics, and related scientific domains.
Built using the Python programming language, NERDD establishes a modular infrastructure specifically tailored for deploying both predictive and generative models.
A central feature of the NERDD framework is its native support for chemistry-specific computational tasks. These chemistry-aware capabilities include native support for molecular property prediction, atom-level inference tasks, and generative workflows focused on molecular generation in drug discovery applications.
To enable the seamless integration of in silico tools, the platform utilizes standardized configuration files alongside containerized modules.
Interaction with the deployed models is supported through two primary mechanisms: a REST application programming interface and an intuitive web interface.
On the backend, NERDD couples a modern architecture with specialized open-source technologies to ensure high availability, fault tolerance, and dynamic scalability. The platform integrates Kafka-based messaging to help ensure scalable, highly available operation. The infrastructure also leverages Ceph to ensure high availability, fault tolerance, and dynamic scalability.
The platform utilizes Kubernetes orchestration to ensure scalable, highly available operation. The platform also employs KEDA-driven autoscaling.
To guarantee robust and reproducible operations, NERDD implements configuration-driven pipelines equipped with automatic input handling. The underlying processing workflow incorporates checkpoints and idempotent handling for robust, reproducible results. As an open-source framework, NERDD empowers researchers in drug discovery to deploy and maintain machine learning models in a reproducible, sustainable manner.
Subject of Research: Microservices platform for deploying machine learning models in cheminformatics
Article Title: The NERDD microservices platform for deploying machine learning models in drug discovery
Article References: Hirte, S., Scholz, V.-A., & Kirchmair, J. (2026). The NERDD microservices platform for deploying machine learning models in drug discovery. Journal of Cheminformatics. https://doi.org/10.1186/s13321-026-01312-4
Image Credits: AI Generated
DOI: 10.1186/s13321-026-01312-4
Keywords: NERDD platform, machine learning deployment, cheminformatics, drug discovery, bioinformatics, microservices, predictive models, generative models, molecular property prediction, atom-level inference, molecular generation, REST API, Apache Kafka, Kubernetes orchestration, KEDA autoscaling, Ceph storage, containerized modules, idempotent processing
Cite Scienmag News
Teresa Odom. (November 18, 2026). Open-Source Platform Deploys Chemistry Machine Learning Models for Drug Discovery. Scienmag. https://scienmag.com/open-source-platform-deploys-chemistry-machine-learning-models-for-drug-discovery/
Teresa Odom. "Open-Source Platform Deploys Chemistry Machine Learning Models for Drug Discovery." Scienmag, 18 November 2026, https://scienmag.com/open-source-platform-deploys-chemistry-machine-learning-models-for-drug-discovery/. Accessed 18 November 2026.
Teresa Odom. "Open-Source Platform Deploys Chemistry Machine Learning Models for Drug Discovery." Scienmag. November 18, 2026. https://scienmag.com/open-source-platform-deploys-chemistry-machine-learning-models-for-drug-discovery/








