GenAIOps

GenAIOps - OpenSearch for AI & GenAI Platform Observability - Ramesh Kumar Manickam & Rama Pabolu, Transunion As Organizations adopt and build out Internal Platforms and Applications that leverage Public and Enterprise GenAI Platforms, It becomes a key imperative to Solidify , Streamline and Protect this base platform infrastructure through Observability that is proactive and GenAI-conversant. In this session , we discover how open source projects like OpenTelemetry, Jaeger ..etc can enhance observability and security for Generative AI (GenAI) infrastructure like Agents, AI components such as VectorDBs, Agents, RAG systems, Inference pipelines, etc., using modern, AI-enabled security guidelines. In this evolving world of AI-driven services , We look at Leveraging OpenSearch and customizations that enable GenAIOps - through end to end AI & GenAI platform observability with necessary support for GenAI Commons driven frameworks like Feature stores (Feast) , Model registries and Execution environments .Challenges and Lesson learnt in tactically and strategically developing, testing and deploying and evolving Application and AI/ Model Monitoring, Inference Monitoring Architectures will be shared that can benefit as best practices broadly across Enterprise GenAI arch community.