1:30pm - 2:30pm (EST) [1hr]
Keynote \ IBM CAS Project Presentation
Room C2
Felipe Rivera, University of Victoria; Hausi A. Müller, University of Victoria; Laura Shwartz, IBM; Ian Watts, IBM
The ever-increasing operational complexity and uncertainty of cloud-based microservice systems challenge human reasoning and problem-resolution abilities. IBM Instana Observability represents significant progress toward improved system understanding for issue resolution. Moreover, integrating observability and digital twins (DTs) for experimentation allow developers to simulate and test diverse scenarios in controlled environments with lessened risk. However, large amounts of data produced in real and simulation environments demand innovative practices to reduce human cognitive load when interpreting results. We discuss the transformative power of large language models in synthesizing actionable insights from knowledge graphs describing relevant historical system states derived from DTs.
2:30pm - 3:00pm (EST) [30 mins]
Demo\Tutorial
Room C2
Felipe Rivera, University of Victoria
This demo presents InstructLab, a groundbreaking platform designed to bring collaborative power to AI model training. By allowing community members to contribute unique skills and domain knowledge, InstructLab enables continual model refinement without the need for extensive technical expertise or costly retraining processes. Participants in this demo will gain hands-on experience with InstructLab’s intuitive interface, learning how to make impactful contributions to large language models (LLMs) in real-time. Through step-by-step interactions, attendees will see how InstructLab’s tools for taxonomy-driven data curation and large-scale synthetic data generation support collective knowledge-sharing and iterative tuning. This session invites users from all backgrounds to explore the democratization of AI, offering a unique opportunity to directly influence the capabilities of advanced AI systems.
Short\Coffee Break (20 mins)
3:20pm - 4:20pm (EST) [1 hr]
Keynote \ IBM CAS Project Presentation
Room C2
Marin Litoiu, York University; Komal Sarda, York University; Ian Watts, IBM
In the dynamic landscape of cloud-based microservice applications, the complexity and interdependencies of system components often lead to intricate operational incidents that challenge conventional diagnostic methods. Continuous operation and minimal downtime necessitate automated detection and remediation of the incidents. This presentation summarises the application of advanced large language models (LLMs) within a proposed robust framework to improve incident management, reduce downtime, and foster a proactive operational environment. The research highlights the potential of LLMs to provide alert summaries with probable root cause detection and remediation script generation, leveraging tools like Ansible and Kubernetes to offer operational resilience and scalable auto-remediation.
4:20 - 5:00pm (PST) [40 mins]
Fishbowl Panel \ Interactive Session
Room C2
Ian Watts, IBM; Hausi A. Müller, University of Victoria, Marin Litoiu, York University; Felipe Rivera, University of Victoria
This engaging panel discussion explores the transformative role of Generative AI (GenAI) as a catalyst in enhancing proactive AIOps capabilities. As cloud operations grow in complexity, traditional methods often fall short of delivering timely insights and proactive solutions. By leveraging GenAI, we open doors to advanced data-driven decision-making, predictive analysis, and optimized problem resolution within AIOps frameworks. The panel will bring together experts from academia and industry to discuss how GenAI enables data to flow into actionable intelligence, supporting both automated and human-in-the-loop decisions. Key topics will include the integration of GenAI with Digital Twins for real-time monitoring, its role in predictive analytics for risk mitigation, and its ability to enhance root cause analysis. Participants will gain insights into the latest advancements, potential challenges, and future directions of GenAI-powered AIOps, setting the stage for operational innovation across complex cloud environments.