Cloud & AI Infrastructure Day 2 Highlights

Scalable AI Infrastructure: Structuring for Innovation

Charles Ewen, Chief Information Officer - Met Office
Leigh Smith, Product Director - AI - Freshworks
Usman Ikhlaq, Programme Manager - Artificial Intelligence - TechUK

As AI adoption accelerates across the UK, this session explored how organisations are preparing for the Agentic AI era. The panel examined what separates experimentation from scalable impact, the remaining barriers, and how a human-centric approach can unlock sustainable transformation.

Micro-Edge Cloud isn’t Coming – It’s Here, Now and Critical

Chairperson: Andy Bates, Chief Information Security Officer - StonesThro

Nicky Stewart, Senior Advisor - Open Cloud Coalition
Diar Abdlkarim, Director - Motion Dynamics
Dr Rois Ni Thuama, Founder - RTConsulting Limited

Hear industry-leading experts debate the opportunities and challenges that micro-edge cloud presents and why your organisation needs to investigate and embrace its potential.

This session brought together leaders from cloud policy, governance, cyber defence and applied AI to go beyond the technology and explore its application.

DigitalOcean and AMD’s Delivery of a 2x Production Inference Performance Increase: A Case Study on Character.ai

Jeff Fan, Senior Solutions Architect - DigitalOcea

Most organizations scale AI inference the obvious way: buy more GPUs. But what if the real bottleneck isn't hardware — it's how you deploy it? In a collaboration with Character.AI (~20M users) and AMD, DigitalOcean achieved a 91% throughput improvement with zero additional spend. The breakthrough came from challenging a default assumption about GPU allocation that most teams never question. This talk distills the technical discovery into a strategic framework that engineering leaders and executives can use to evaluate their own AI infrastructure investments. You'll leave with a decision matrix, not a shopping list.

How Open Models Transform AI Economics

Seif Mazigh, Head of European Solution Engineering - Crusoe

Facing soaring AI costs and a lack of control? Hear why embracing open models is the only way to secure IP and gain model control. Learn how Crusoe's vertically integrated infrastructure are engineered to drastically lower your cost-per-token and achieve predictable, profitable enterprise-scale AI deployment.

Be More Valuable. Build Your Human Skills That AI Can't Replace

Ben Pearce, Founder - Elevated You

AI just can't replace some skills. So those skills become more valuable. In this interactive session Ben spoke about the essential human skills that make you more valuable in the tech world. Skills like delivering presentations, influencing stakeholders, giving feedback, active listening, having tough conversations.

Fleet at Planet Scale: Challenges in Managing Millions of Bare-Metal Servers

Igor Kanyuka, Staff Software Engineer - Meta

At planet scale, managing millions of bare‑metal servers requires end‑to‑end, automation‑first lifecycle management. This talk surveys the core challenges—heterogeneous hardware, high‑throughput provisioning, rapid error detection and remediation, safe kernel/firmware/OS maintenance, and sustainable decommissioning. The key message: at planet scale, reliability is engineered through automation, observability, and governance, turning manual operations into predictable, auditable, self‑healing processes.

Agents at Scale: Treating Context as Infrastructure

Vishal Bala, Senior Applied AI Engineer - Redis

Building an agent demo is straightforward. Running agents reliably in production is a different challenge. As usage grows, the hard problems show up around context: keeping it fresh, retrieving it quickly, isolating it per user, and ensuring it doesn’t overwhelm the model or your infrastructure. This talk walks through architectural patterns for building agents that scale, make effective use of context, and manage and persist memory for dynamic and personalized experiences.

Design Decisions that Shape Load and Traffic: A Case Study of a Search API

Ellie Rahimi, Senior Software Engineer - Trainline

This case-study-driven talk uncovers how small, often overlooked design decisions can lead to significant changes in service traffic and customer experience. We will discuss how "design drifts" in internal service calls can quietly increase traffic, cost, and risk. Ellie shared a production case study where reclassifying requests, moving lookups to authoritative domains, and tightening contracts reduced total traffic by approximately 92%. This session demonstrates how early design choices ripple through a system, shaping performance and user impact in ways that may not be visible until much later.

Strategic AI: transforming investment into measurable AI

Brian Horsburgh, UK SMB Director - DELL TECHNOLOGIES

Unlock the true value of AI by turning strategic investment into tangible business outcomes. In this session discover how to maximise AI investments on a budge, aligning them with clear achievable metrics that track progress and impact. Learn how to integrate AI into your business strategy to boost productivity accelerate growth and maintain a competitive edge. Whether you're navigating early adoption or scaling existing initiatives, this talk will provide actionable insights to help you make smarter more measurable AI decisions.

Re-Engineering Corporate Engagement: A New Model for Scaling Enterprise Growth

Deryk Williams, Managing director - Inception Outreach

In this session, Deryk outlined Inception Outreach's unique engagement model and commercial approach, sharing how it has supported the creation of over £800m in pipeline over the past three years. Attendees gained insight into how rethinking engagement at scale can unlock growth, improve market access, and deliver measurable commercial impact in competitive technology environments.
 

AI and Sovereignty in AWS

Muhammad Basit, AWS Sales Lead UKI - Noventiq

Noventiq’s session on AI and Sovereignty in AWS explores how organisations can harness cutting-edge AWS artificial intelligence capabilities while ensuring data governance, sovereignty, and regulatory compliance across borders. Backed by Noventiq’s deep AWS expertise and strategic collaboration with Amazon Web Services, the talk highlights practical strategies for secure, sovereign AI deployment in the cloud, real-world use cases, and best practices. Presented by Muhammad Basit, this session delivers actionable insights for leaders driving cloud-native innovation.

The Death of the Dashboard: How Observability and AI Are Redefining IT Operations

Mark Roberts, CEO - Prosperon Networks / SolarWinds

This session explored the evolution from monitoring to observability, and onwards to AI-driven operations, and how the next five years will fundamentally change the human role in IT. Rather than analysing data and reacting to incidents, humans will increasingly act as reviewers and governors, validating automated preventative actions, incident responses, and resolutions driven by AI-powered automation engines.