Most enterprise AI coding pilots stall somewhere between the demo and production. I'm a Solutions Engineer for Cloud & AI at Microsoft, and closing that gap is the job. This past year that's meant taking one customer's GitHub Copilot footprint from 8,000 to 26,000 developers, and running 10+ hackathons and 50+ workshops to make sure the licences turned into working practice rather than shelfware. Before this, nearly seven years at Red Hat. I built APAC's first ROSA architecture from scratch which now is the reference pattern used across the world. I replatformed one of India's largest private hospital chains from VMware to OpenShift Virtualization and cut their infrastructure cost 80%. I co-owned 180 corporate accounts and co-led the national roadmap for Ansible adoption. Day to day I sit inside the customer's environment and remove whatever is blocking production: integration with legacy systems and identity, data readiness, agent orchestration and state management, evaluation harnesses, token cost and latency. Architecture on a slide is easy. Shipping it alongside the customer's own engineers is the work that counts. In 2022, Opensource.com readers voted my Kubernetes writing the People's Choice community award. I write here about what happens when AI agents meet real codebases: isolated worktrees, merge strategy, review load, the parts the demos skip. About 12,000 engineers follow along. Off the clock I build cloud-native communities, and practise yoga and meditation with the Art of Living Foundation.