I’m beyond excited to share my top picks for “5 KubeCon Talks Not to Be Missed” coming up at KubeCon + CloudNativeCon India 2025, happening in Hyderabad, India, on August 6–7. This is the second time KubeCon is happening in India, and the vibe just keeps getting better. Last year was amazing, and I can already tell this year will be even more inspiring.
In this blog, I’ve combed through the agenda and handpicked the five talks that you should attend and I’m most looking forward to.
1. Kubernetes for GenAI: The New OS of the AI‑First World
Janakiram MSV, Principal Analyst at Janakiram & Associates
A keynote exploring why Kubernetes is emerging as the foundational platform for generative AI workloads. You can expect insights on optimizing clusters for AI, model orchestration, and real-world implementations in AI deployments, perfect for readers interested in cloud-native AI evolution.
2. Pushing the Limits: Scaling ArgoCD’s Repo‑Server To Handle 30K Applications
Aditya Sharma & Isha Gusai, Expedia Group
This talk explores the ArgoCD scalability challenges and addresses real-world pain points around managing thousands of deployments. You can expect technical strategies, architectural optimizations, and practical lessons, cool for teams scaling GitOps at enterprise levels.
3. Handling Node Churn in Karpenter: Efficient Scaling for Large EKS Clusters
Shivani Mehrotra (Expedia Group) & Chetan Saini (Antra)
In this talk, you will learn practical approaches to analyze and reduce node churn in EKS using Karpenter, including memory metrics, PDBs, and Pod Priority strategies. Expect insights on optimizing cost and performance by blending Spot, Reserved, and On‑Demand EC2 instances, fine‑tuning node expiration settings, and utilizing Karpenter’s latest features like consolidation and do‑not‑disrupt annotations.
4. Predictive Autoscaling in Kubernetes With KEDA and Prophet
Snigdha Kanchana, IBM
In this talk, you’ll learn how to implement predictive autoscaling by combining KEDA with Prophet to forecast workload trends and initiate scaling before traffic spikes occur, resulting in lower latency and smarter resource usage.
5. Scaling AI Like a Pro- PepsiCo’s LLM Deployment Strategy on Kubernetes
Praseed Naduvath & Dhanashree Shetty, PepsiCo
In this talk, you will learn how PepsiCo is deploying large language models at scale within Kubernetes. You will learn about architectural decisions, GPU provisioning strategies, and fine-tuning techniques. This is great for those interested in hybrid workloads, ML operations, and how major enterprises are integrating LLMs into production.
Come and Meet the PerfectScale Team at Booth S12
Stop by our booth in Hyderabad on August 6–7. We’re excited to chat in person about how to optimize Kubernetes, boost reliability and efficiency, and help you cut cloud costs by up to 50%. Whether you're running EKS, GKE, or AKS, we'll show you how PerfectScale provides real-time visibility, autonomous right-sizing, and intelligent scaling across multi‑cloud, multi‑cluster environments. Our team will be on hand to answer your questions, walk you through live demos, and yes, give you some cute stickers to take home. Can’t wait to meet you there!
.png)
