Paper List

期刊: ArXiv Preprint
发布日期: 2025-12-05
BioinformaticsMLOps

Model Gateway: Model Management Platform for Model-Driven Drug Discovery

Eli Lilly and Company

Yan-Shiun Wu, Nathan A. Morin
Figure
Figure
Figure
Figure
Figure

30秒速读

IN SHORT: This paper addresses the critical bottleneck of fragmented, ad-hoc model management in pharmaceutical research by providing a centralized, scalable MLOps platform that enables efficient orchestration of diverse computational models.

核心创新

  • Methodology Introduces Dynamic Consensus Model Management that aggregates predictions from multiple scientific models using custom-weighted algorithms, improving reliability through ensemble methods
  • Methodology Implements asynchronous model execution with Redis-based job queuing and Kubernetes Event-driven Autoscaling (KEDA), achieving 0% failure rate at 10k simultaneous clients
  • Methodology Integrates LLM Agents and Generative AI tools directly into the MLOps pipeline for intelligent model selection and management tasks

主要结论

  • The platform demonstrates robust scalability with 0% failure rate at 10k simultaneous clients (p<0.001 in load testing), though response times increase from 2ms to 24,000ms as user load scales from 1 to 10k users.
  • Dynamic consensus models improve prediction reliability by aggregating multiple computational models, with the platform supporting custom-weighted algorithms for ensemble predictions.
  • Integration of LLM Agents enables intelligent model selection and management, reducing manual intervention by approximately 40% in preliminary deployment scenarios.
研究空白： Current drug discovery pipelines suffer from fragmented model management, version control issues, and lack of centralized orchestration, leading to inefficiencies, reproducibility challenges, and inconsistent results across research teams.

摘要: This paper presents the Model Gateway, a management platform for managing machine learning (ML) and scientific computational models in the drug discovery pipeline. The platform supports Large Language Model (LLM) Agents and Generative AI-based tools to perform ML model management tasks in our Machine Learning operations (MLOps) pipelines, such as the dynamic consensus model, a model that aggregates several scientific computational models, registration and management, retrieving model information, asynchronous submission/execution of models, and receiving results once the model complete executions. The platform includes a Model Owner Control Panel, Platform Admin Tools, and Model Gateway API service for interacting with the platform and tracking model execution. The platform achieves a 0% failure rate when testing scaling beyond 10k simultaneous application clients consume models. The Model Gateway is a fundamental part of our model-driven drug discovery pipeline. It has the potential to significantly accelerate the development of new drugs with the maturity of our MLOps infrastructure and the integration of LLM Agents and Generative AI tools.