Josh Bayne
Founding Engineer | AWS AI League Champion London 2026
Josh Bayne is an AI Engineer at Maiven with a background in data engineering and a Master’s degree in Physics, working on LLM systems, agent orchestration, and production AI tooling. His work focuses on practical approaches to building AI systems that are reliable, observable, and easier to reason about in real-world environments, with a strong emphasis on experimentation, measurement, and keeping complexity under control.
Josh was the winner of the AWS AI League London finals, where he built an autonomous AI agent using Amazon Bedrock and AgentCore capable of navigating dynamic environments, solving coding and logic tasks, and adapting under time pressure. His approach focused on simple, reactive system design rather than heavily layered prompting or orchestration.
Outside of AI, Josh enjoys video games, good food (sometimes making, mostly eating), and spending time on creative side projects and experiments.
"You Got This": Why Simple Agents Win
The more dynamic the environment, the more boring your orchestration should be...
In the AWS AI League London finals, the winning edge was not an elaborate prompt stack or a swarm of agents. It was a simple architecture: explicit state, narrow tools, reactive logic, and orchestration simple enough that failures were obvious. This talk argues that the KISS principle is not anti-agent; it is how you build agents that survive real environments. I’ll show where LLM reasoning helps, where deterministic software should take over, and how to recognise when added complexity is actually justified.