I design systems, run infrastructure, and finish what I start.
Twenty years of hands-on IT — user support, systems administration, networks, cloud and security — and more recently two cloud platforms I designed and built myself. This site shows the work: the constraints, the decisions and the trade-offs, so you can judge it for yourself.
Case studies
Two products I designed, built and run myself — written up honestly, with the constraints and trade-offs spelled out.
CaseConduit.io — from blank page to a live, billing SaaS platform, solo
An AI-powered case-method teaching suite on a fully serverless AWS architecture: one governed chokepoint for auth, metering, cost logging and billing; a resume protocol that makes gateway timeouts invisible; pricing derived from measured AI cost; SSO and LTI 1.3 for institutional buyers.
Read the case study → Product build · 2026UpConduit — a global monitoring SaaS in 48 hours
Uptime and email-health monitoring from a seven-region serverless probe network: flap-resistant alerting, canary-validated blocklists, heartbeat monitoring, Stripe billing, public status pages and a live world map. Built quickly — largely because the patterns were already proven on CaseConduit.
Read the case study →Work experience
Roles for employers and clients — systems administration, networks, and end-user IT support, which I've done throughout my career and still enjoy.
SEA Construction — retiring a legacy platform without drama
Replacing a construction manufacturer's ageing ASP.NET site and custom CMS with a static-first AWS architecture: S3 and CloudFront hosting, a draft-and-publish content pipeline, and an admin portal designed around Cognito — planned as incremental passes a small business can absorb.
Read the case study → The long haul · 2009 – 2025The Case Centre — fifteen years of quiet infrastructure
Sole IT administrator for an international publisher: migrating on-premise servers to AWS, moving the organisation to Microsoft 365, building the disaster-recovery plan, achieving Cyber Essentials, and running a £100K annual budget — the long-haul version of the same discipline.
Read the case study →Small, on purpose
Right-sizing is a skill too. Not every problem deserves an architecture.
Commission · 2026jlbrooke-author.co.uk — an author's home on the web
A crime and horror novelist needed a professional presence: book, buy link, press images, socials, contact. Delivered as a single hand-built page — no CMS, no framework, nothing to maintain — on the same S3 and CloudFront foundation as everything else I run, with search and social-sharing metadata done properly. Costs a few pounds a year and will outlive every platform trend.
Visit the live site →How I work
Kaizen, genuinely.
I ship in small, complete passes. Each pass leaves the system live, coherent and better than the last — the current product is always the new baseline, never a building site. It's how one person maintains production systems without heroics, and it's how I've worked for twenty years.
Support is part of the job — and I like it.
Most of my career has been front-line IT: helpdesks, desk-side fixes, being the person people call when their day stops working. I still enjoy that work. Supporting real users keeps you honest about how systems actually get used, and it's where I learned to explain technical things in plain English. Systems administration and user support aren't what I do between projects — they're the foundation everything else here is built on.
Constraints before solutions.
Every design on this site started from its constraints — a solo operator, a hard platform limit, a marginal cost per click — because a solution that ignores its constraints is a demo. The interesting engineering is usually in what you decide not to fight.
Honest about AI.
I build with Anthropic's Claude as a force multiplier and say so plainly. My value is the layer AI doesn't provide: the architecture, the failure-mode thinking, the security posture, the economics, and the judgement about what's worth building. Direction is a skill; twenty years of running infrastructure is where mine comes from.
Boring is a feature.
Static files, managed services, one chokepoint, alarms that only fire when something is actually wrong. I optimise for systems that let their operator sleep — including when the operator is me.