KubeCon + CloudNativeCon India 2025: Technical Insights from Hyderabad

Introduction

Hyderabad welcomed KubeCon attendees on 5th August with monsoon rains and the familiar energy of thousands of engineers converging to push cloud native innovation forward. Walking into the venue, the scale became immediately apparent as developers, platform engineers, and community members gathered to exchange knowledge, drink chai, and tackle the evolving challenges of running Kubernetes at scale.

At LiveWyer, we’ve been tracking the platform engineering trends emerging from these conferences, and this year’s Indian edition provided valuable insights into how security, scaling, and governance are evolving in production Kubernetes environments.

Three sessions that changed our perspective

Handling Node Churn in Karpenter: Efficient Scaling for Large EKS Clusters

Speakers: Shivani Mehrotra & Chetan Saini (Expedia Group)

Shivani Mehrotra and Chetan Saini from Expedia Group tackled node churn, that constant cycle of nodes spinning up and terminating in large clusters. Using Karpenter, they demonstrated strategies for reducing churn through pod priorities, pod disruption budgets, and intelligent blending of spot, reserved, and on-demand instances.

The cost optimisation impressed us, though mixing instance types effectively can deliver significant savings. But Karpenter’s flexibility in launching exactly the right instance type at the right time caught our attention more. This level of control translates to better application stability and more predictable cluster behaviour.

My takeaway:

Our experience with large-scale EKS deployments echoes their findings. Scaling means making intelligent decisions that balance cost, performance, and reliability.

Karpenter Documentation | EKS Best Practices

Guard your Network with Kyverno and Envoy

Speakers: Sanskar Gurdasani (AccuKnox) & Swastik Gour (InfraCloud Technologies)

Sanskar Gurdasani from AccuKnox and Swastik Gour from InfraCloud Technologies demonstrated how access control can be applied at the cluster level using Common Expression Language (CEL) through the Kyverno Envoy Plugin. Their approach requires zero application code changes.

This integrates naturally with service meshes like Istio, embedding security controls into the platform itself. We’ve seen similar patterns work well in our client environments. When security becomes invisible to developers, they’re more confident shipping features quickly.

My takeaway:

The architectural insight here transforms future conversations. Instead of asking developers to implement security controls, the cluster handles them automatically. This shifts focus from avoiding mistakes to building with confidence.

Kyverno Policy as Code | Envoy Proxy

Kubernetes Policy as Code (PaC) for Platform Engineers

Speakers: Sonali Srivastava (InfraCloud Technologies), Mohd Kamaal & Kushal Agrawal (Independent)

Sonali Srivastava from InfraCloud Technologies, along with Mohd Kamaal and Kushal Agrawal, explored Policy as Code using Kyverno. Policies become versioned, consistent, testable code that ensure compliance requirements can be automatically enforced rather than manual processes that can potentially lead to inconsistent policies across environments.

This resonates with our platform engineering work. Manual processes don’t scale with team growth, and Policy as Code ensures consistent application of rules across environments. We’ve implemented similar approaches where policies are treated as infrastructure, reviewed, tested, and deployed through the same pipelines as application code.

My takeaway:

The practical benefit is immediate. Developers get fast feedback on policy violations during development, while platform teams maintain consistent governance without manual intervention.

Kyverno Policy as Code | CNCF Blog on PaC

Practical insights for Platform Engineers

Security as a Platform Feature

The Kyverno sessions highlighted a shift we’re seeing across client projects. Security enforcement is moving from the application layer to the platform layer. When policy violations are caught in CI/CD pipelines, developers receive immediate feedback without needing to understand the underlying security frameworks.

This approach has proven particularly effective in multi-team environments. The platform embeds security practices automatically rather than training every developer on security frameworks.

Cost Optimisation through Intelligence

Both the Karpenter session and other scaling-focused talks emphasised modern autoscaling through adding the right resources. The combination of workload analysis, instance type selection, and capacity planning can dramatically reduce cloud spend while improving reliability.

Our client work often begins with clusters that scale reactively. Moving to predictive scaling patterns, supported by tools like Karpenter, typically reduces costs by 30-40% while improving application stability.

Platform Engineering fundamentals

A conversation with David O’Dwyer, LiveWyer CEO, reinforced something we emphasise with all client teams. Engineering work extends beyond immediate functionality. Code must be readable by the entire team, technical debt should be managed proactively, and documentation needs to be treated as a deliverable.

This mindset becomes critical as Kubernetes environments grow in complexity. Platform teams that invest in clear documentation and knowledge sharing scale more effectively than those focusing solely on technical solutions.

The broader Platform Engineering picture

Three themes emerged consistently across multiple sessions. Automation reducing manual effort, policies becoming code, and security becoming a platform concern rather than an application concern.

We’re seeing these patterns in our client work as well. Organisations that treat their Kubernetes platforms as products, with clear interfaces, consistent policies, and automated operations, deliver features faster and more reliably than those managing infrastructure reactively.

The technology choices matter, but the approach matters more. Tools like Karpenter, Kyverno, and other policy engines succeed because they align with how teams actually work.

Looking forward & final takeaways

KubeCon India 2025 highlighted that successful platform engineering creates systems that help teams ship software confidently and efficiently. The conversations, both in sessions and hallway discussions, consistently returned to making developers’ lives easier while maintaining operational excellence.

Final takeaways from me:

• The cloud native ecosystem continues to evolve rapidly, but the core principles remain consistent. Automate repetitive tasks, make the right choice the easy choice, and design platforms that scale with your organisation.

• Platform teams evaluating these tools should ask whether they align with how your teams want to work tomorrow, beyond solving today’s problems.