Scaling your Cloud Platform Without Losing Cost Control

Cloud costs spiraling? At Rokt, platform engineering became a product with shared ownership, unit economics and embedded FinOps. Learn how we tied platform efficiency to business outcomes and scaled Kubernetes without runaway spend.

No More War Rooms: Kubernetes MCP Servers + Agentic AI for Autonomous Multi-Cluster Remediation

What if your on-call never had to touch a cluster again? This talk demonstrates an agentic AI system combining a Kubernetes MCP server layer across multiple clusters with an LLM orchestration loop to detect, diagnose and remediate issues autonomously. We cover the hub-and-spoke MCP architecture, agent guardrails, and real lessons from running this at scale

Hands-Off Rightsizing: How We Tuned Spark Resource Requests Automatically

Rokt automated resource optimization across 1,000+ daily Spark jobs running on Amazon EKS. The talk covers the JVM visibility gap that makes Spark uniquely difficult to rightsize, Rokt's graduated rollout approach using ScaleOps, and the real-world results: 30–35% reduction in CPU and memory requests. Hear from a hands-on experience our practical gotchas around policy segmentation, data skew, and Spot instance compounding.

Slide Deck: https://docs.google.com/presentation/d/1xYAY9T7roBzejfxc9bbGVz6QBRggguCvOypwFzPaJX0

Meetup Link: https://luma.com/p18cq1vz

How to stop Over provisioning Kubernetes resources

Kubernetes workload optimisation, whether through moving to Amazon Web Services (AWS) Graviton instances and leveraging new scaling solutions such as Karpenter. In today's digital landscape, where workload demands fluctuate with seasonality and regional differences, manual scaling is no longer sustainable.
Join me at AWS Sydney Summit (2025) on a level 400 deep dive into how they implemented and continue to iterate on automated rightsizing strategies across their global infrastructure.

https://aws.amazon.com/events/summits/sydney