Senior Rails consultant with an AI-native delivery pipeline. Critical systems kept reliable — and moved forward at a pace that used to take a team.
Get in touchSenior Ruby on Rails developer and GIS/automation specialist, available fractionally or as a freelancer — experienced, independent help without agency overhead.
I specialise in the harder end of Rails work: legacy upgrades, long-running system stability, deployment and ops, and the operational problems that don't have obvious answers. A particular niche in geospatial systems — map platforms, planning and property portals, FME/Rails APIs, CAD/web integrations and geospatial automation.
What has changed is the throughput. Every part of my delivery pipeline is now agentic: implementation written in Cursor with Grok's Composer model, every pull request reviewed automatically by Copilot before I look at it, and Sentry wired into Slack so production tells us what's actually broken. The judgement about what to change — and what not to touch — is still mine. The tedious mechanical work is not.
Need experienced Rails help without hiring full-time? Get in touch.
AI isn't a line on my invoice — it's how the work gets done
I write the spec, the constraints and the failing test. The agent works to them. Unbounded prompting is how AI-assisted projects go wrong.
The bulk of implementation is agent-written, in-repo, with the whole codebase as context — not copy-pasted from a chat window.
Automated review runs on every change before a human sees it. Human review time then goes on design decisions instead of typos and missed edge cases.
Agentic speed is only safe with a suite that fails loudly. On legacy codebases, coverage comes before velocity — every time.
Production errors arrive in Slack with context, get triaged the same day, and are often fixed before anyone raises a ticket.
Deploys, error rates and what changed, reported into your Slack. You shouldn't have to ask a contractor what happened this week.
Typical clients who hire me
Things I do for clients
Building from scratch, or adding features to existing systems — agent-written, reviewed twice, shipped behind tests.
Rails majors, Ruby versions, dependency debt. Agentic tooling collapses the mechanical part of an upgrade; the judgement about what will break is the part you're paying for.
Software is never "done" — libraries and languages get updated, operating systems get patched, servers become outdated. I keep you on supported versions before it turns into an emergency.
Working with your team to decide what's worth building — including where AI genuinely helps and where it's a liability.
Setting your developers up with agentic coding, automated review and error-to-agent loops — plus the guard rails that stop it going wrong.
AI-assisted tools for the workflows your team does by hand: review, validation, data wrangling, reporting.
Minimising downtime, outages and flaky tests. Production always tells you what's wrong — the job is making sure something is listening.
Success is great, but will the site stay up if you double your users? Triple? ×100?
Sentry, structured logging and alerting wired into Slack. You should be the first to know about problems — ideally alongside the fix.
Rigorous tests are what make fast, AI-assisted delivery safe rather than reckless. They're the foundation, not the afterthought.
A selection of recent client work
Tools and platforms I work with
Day rates for UK senior Rails work — you're buying judgement and accountability; the throughput comes with it
Available for upgrades, troubleshooting, architecture, automation and operational support — and for helping your team adopt agentic development.