Skip to content

About us

Cloud Native London is a strong, open, diverse developer community around cloud, data and AI in London. Since 2017, our monthly events have brought together 10k members, primarily senior DevOps / platform engineers and technical architects.

Speak or sponsor: http://cloudnativelon.com/

All members are required to agree with the Community Code of Conduct.

Sponsors: TechShack are a tech recruitment agency specialising in the Cloud, Engineering and AI. TechShack exist to build high-performing tech teams by putting community, connection, and quality first. Our vision is to create the most trusted tech recruitment community, where meaningful connections drive innovation and growth.

Cloud Native London, November 2026

Cloud Native London, November 2026

Just Eat London, Fleet Place House, 2 Fleet Pl, London EC4M 7RF,, London, GB

Hi folks!

Welcome to our November Cloud Native London meetup! Join us to hear from our three great speakers and network with your fellow techies over pizza and drinks, or alternatively chat and following along on Youtube or LinkedIn!

6:00 Pizza and drinks
6:30 Welcome
6:45 Admission Control for AI Coding Agents: Fail-Closed Policy on Every Tool Call (Ace Vinayak, Pluto)
7:15 Kubernetes Is Not Your Hyperledger Fabric Control Plane (Pereowei Daniel, DLM Capital Group)
7:45 Break
8:00 RabbitMQ cluster migration (Surya Pandian, Truelayer)
8:30 Wrap up

See you there!

Cheryl (@oicheryl)

Admission Control for AI Coding Agents: Fail-Closed Policy on Every Tool Call (Ace Vinayak, Pluto)
Coding agents now run kubectl, terraform, git and cloud CLIs with an engineer's own credentials. We already know how to stop a bad manifest reaching a cluster: an admission controller checks it against policy and rejects it by default. Agents have no equivalent. This talk applies the same idea one step earlier: turn each agent tool call into facts (what it runs, what it touches, which environment), decide it against written policy deterministically, and fail closed when a call can't be understood.
I'll cover the controls that broke first (sandboxes, allowlists, hooks, one LLM reviewing another), the bypasses we found in our own gate and how we closed them, and allow / ask / block tiers that avoid approval fatigue. Policies live as code with regression tests, like any other cluster policy. Vendor-neutral: the pattern works with any agent and any policy engine."

Ace Vinayak is the founder of Pluto- We build Warden, a coding agent that checks every tool call an AI agent makes against written policy before it runs, failing closed by default. He holds a PhD in computer science from the University of Surrey and was previously co-founder and CTO of PerchPeek, where he scaled a B2B platform to enterprise customers across the UK, US and EMEA.
LinkedIn: linkedin.com/in/aasisvinayak

Kubernetes Is Not Your Hyperledger Fabric Control Plane (Pereowei Daniel, DLM Capital Group)
Kubernetes can schedule Hyperledger Fabric peers, orderers, certificate authorities, and chaincode services, but a running Pod does not necessarily mean that a Fabric network is operational. Kubernetes cannot tell whether a peer has the correct MSP material, whether certificates are valid, whether it has joined the intended channel, or whether a Chaincode-as-a-Service deployment is ready to serve transactions.
This talk explores the gap between Kubernetes workload health and Fabric operational readiness. Drawing on the development of FabricOps, an LF Decentralized Trust Lab, I will show how a reconciliation-driven Kubernetes operator can express Fabric networks as desired state and manage the lifecycle concerns that matter in production: identities, TLS, channels, chaincode, health, and multi-organization ownership. The session will include a short demonstration and an honest look at the remaining hard problems, including federated network joining and BFT ordering.

Pereowei Daniel is a DevOps/Cloud Infrastructure Engineer, and the maintainer of FabricOps, an LF Decentralized Trust Lab for declaratively provisioning and operating Hyperledger Fabric networks on Kubernetes. He is also a contributor to Hyperledger Labs Fablo, where he contributed Chaincode-as-a-Service support. His work focuses on making the operational complexity of distributed systems, identity infrastructure, and permissioned ledgers more manageable through automation.
Social links:
LinkedIn:
https://linkedin.com/in/pereowei
X: https://x.com/pereoweidaniel
GitHub: https://github.com/dpereowei

RabbitMQ cluster migration (Surya Pandian, Truelayer)
What happens when one message broker cluster sits under every payment your platform handles, and it starts to wobble? Mirrored queues stall, retries loop forever, and one noisy service takes down everyone else.
This talk is the story of how we split a single RabbitMQ cluster into many, moved to quorum queues with delivery limits and dead-letter queues, and did it with zero downtime. You'll learn the three-step ""dual consume"" pattern, the library and tooling traps that bit us, and the ""scream test"" we used to find forgotten users.

Surya Pandian is a Software Developer at TrueLayer, the open banking payments network. She works on the platform's messaging infrastructure and the Rust libraries that sit on top of it. She has led parts of a multi-cluster RabbitMQ migration and fixed a long-standing reconnect bug in the company's Rust messaging libraries. She previously spoke at Cloud Native about migrating a Kubernetes cluster. She's passionate about reliability, open source and making risky migrations boring. Find her on GitHub at github.com/suryapandian.

Check out https://www.oicheryl.com/cloudnativelondon if you're interested in speaking or sponsoring.

  • Photo of the user
  • Photo of the user
  • Photo of the user
142 attendees

Upcoming events

4

See all

Group links

Organizers

Cheryl H. is a Super Organizer

Members

10,842
See all

Find us also at