The Deployment Decade
The hard part of AI stopped being invention and became deployment. What that means for anyone running a transformation, and why the next ten years reward the people who are good at the last mile.
Essays and opinions on AI and what it touches: finance, energy, the shape of organisations, and the world the next generation inherits. Written with optimism, and with evidence.
The hard part of AI stopped being invention and became deployment. What that means for anyone running a transformation, and why the next ten years reward the people who are good at the last mile.
Eight years inside a bank showed me what moves an AI initiative: a data contract somebody owns, and a clear answer to who is allowed to act on the output.
Notes from moving from enterprise software to physical AI. Retry stops being free, and dust beats accuracy.
Written in December 2025, from notes on work between 2016 and 2025.
The pilots that reach production have one thing in common: they touch real production data from the first week. They treat AI as a business capability rather than a tech demo, and they put the "boring" infrastructure in place early: data governance, CI/CD for models, and a clear path to ROI.
Read ArticleWhy Europe's banking sector is moving towards owned infrastructure. With regulatory pressure (GDPR, EU AI Act) mounting and data privacy becoming a board-level risk, relying solely on US-based hyperscalers is no longer a viable long-term strategy for critical infrastructure.
Read ArticleHow to filter the noise of new model releases and focus on what actually moves the needle. Every week there's a new SOTA model on benchmarks that don't matter to your business. This framework provides a scorecard for evaluating new GenAI capabilities against specific constraints: Innovation, Feasibility, and Viability.
Read ArticleAdapting high-velocity physics models for financial fraud detection. Traditional rule-based systems generate too many false positives, annoying customers and burying analysts. We took a different approach, borrowing Markov chain methods from particle physics to detect "interesting events" in noise.
Read ArticleWhat it takes to run a Kafka architecture at billions of events a day. Scaling beyond "Hello World" is where the real engineering begins: keeping a stateful cluster healthy through peak loads, riding out rebalancing storms, and building idempotent consumers.
Read ArticleArchitecting low-latency inference systems years before ChatGPT. In 2018, real-time inference was still unusual outside card fraud and ad tech. We explored the trade-offs between batch and real-time inference, eventually building a custom serving layer using C++ bindings for Python models.
Read ArticleExplorations into autonomous coding agents and their impact on developer velocity. I am currently running experiments to see if an AI agent can build a production-ready SaaS MVP in under 48 hours without human intervention. The results are complex, and I am writing them up as they come in.
Read Article