LOC: Chelmsford, UK ROLE: AI Enablement Engineer STATUS: Open to opportunities

I build AI systems people actually use.

Production agents, LLM pipelines and automation — plus the enablement work that makes them stick: playbooks, hackathons, governance and proper handover. Engineering and adoption, one person.

View deployments Get in touch
Scroll
LLM Triage in Production Agent Orchestration Python · LangGraph · n8n Human-in-the-Loop by Design Executive AI Coaching Weekly AI Hackathons RAG & Evaluation Harnesses Power Platform Delivery Documentation-First Handover
01

Mission Brief

// who I am

Most AI projects don't fail on the model. They fail on adoption. I own both halves — I ship the system, then I make the organisation actually use it.

At Ground Control I lead AI enablement: I built and operate an LLM-powered service-desk platform end to end — triage, reply drafting, Jira integration, human-in-the-loop review, full audit trail — and instrumented it so its value is measured, not claimed.

Around the engineering I run the adoption machine: executive AI playbooks for CIO and director-level leadership, company-wide hackathons, governance and guardrails, and documentation-first handover. Small, boring automations with measurable outcomes beat flashy demos — every time.

Fourteen years leading hospitality teams of up to 50 with full P&L taught me the part most engineers skip: change lands through people, not features.

P.01

Default to robust

Edge cases, retries and failure modes are part of "done" — the happy path is a prototype.

P.02

Measure or it didn't happen

Hours saved, tickets deflected, manual work replaced. No vanity metrics.

P.03

Build to embed, not to own

Nothing ships without documentation and handover notes. Hard line.

P.04

Adoption is engineering

Speed without change management is slow. Training and governance ship with the code.

0+
Years shipping production AI
0
People led at peak, full P&L
0%
NPS on mystery-diner measures
0yrs
Leadership before tech
02

Deployments

// selected work, all in production or live pilot
D.01LIVE

AI Service-Desk Platform

An LLM-based platform that classifies inbound messages, drafts replies and raises Jira issues — orchestrated in n8n on a Supabase/Postgres backend, with a human-in-the-loop reviewer workflow and a full audit trail. Instrumented for message volume, reviewer performance and time saved, then hardened with data-cleanup to close every logging gap.

> production, not demo:
measured triage time replaced,
auditable end to end
LLM Triagen8nSupabaseJira APIHITL
D.02LIVE

Document Intelligence Agent

A Python agent that extracts structured planning data from historical engineering inspection PDFs — turning a paper archive into queryable data. Built alongside a Jira priority-classification automation that routes work before a human ever touches it.

> unstructured archive →
structured, queryable records
PythonPydanticClaude CodeExtraction
D.03INFRA

Unified LLM Gateway

Migrated the organisation's internal AI tooling onto a single OpenRouter API gateway with model fallbacks — standardising access, centralising cost control, and decoupling every project from any one model provider.

> one gateway, every project:
fallbacks + cost control built in
OpenRouterModel RoutingCost Governance
D.04PROGRAMME

AI Enablement Programme

The adoption engine: executive AI playbooks for CIO and director-level leadership, weekly company-wide hackathons, train-the-trainer sessions and shared patterns so non-specialist teams adopt AI safely. Contributor to the internal Copilot adopters newsletter.

> from "AI curious" to
self-serve, governed adoption
Exec CoachingHackathonsGovernanceTraining
D.05DELIVERED

Power Platform Portfolio

Power Apps, Power Automate, Dataverse and Copilot Studio solutions for asset management and new-starter onboarding — every one handed over to its owning team with full documentation. Built to embed, not to babysit.

> handed over + documented —
zero long-term babysitting
Power AppsDataverseCopilot StudioALM
03

Systems & Stack

// daily drivers

AI / LLM Engineering

  • Python — pydantic, SQLModel
  • Anthropic & OpenAI APIs, OpenRouter
  • RAG & agent orchestration
  • Evaluation harnesses, prompt engineering
  • Claude Code, MCP

Automation & Workflow

  • n8n — production orchestration
  • Power Automate & Power Apps
  • Copilot Studio, Dataverse
  • API integration, webhooks
  • Jira & Azure DevOps automation

Data & Platform

  • SQL Server — spatial, stored procs
  • Supabase / Postgres, SQLite
  • OData / Web API, TypeScript
  • Microsoft 365 ecosystem
  • OAuth, key vaults — no hardcoded secrets, ever

Enablement

  • Internal AI tooling & governance
  • Executive coaching & playbooks
  • Train-the-trainer programmes
  • Technical documentation & handover
  • Change management
04

Track Record

// trajectory
2023 — Present

Automation & AI Enablement Lead · Ground Control

Own the organisation's applied-AI portfolio end to end — from production LLM systems to executive coaching and company-wide enablement. The hybrid role: engineer who ships, enabler who embeds.

2008 — 2022

Hospitality Leadership · GM & Management Roles

Led teams of up to 50 with full P&L at premium sites turning over £50–60k a week — Miller & Carter, Galvin Green Man (Michelin Bib Gourmand, Pub of the Year), Chop Bloc, Loch Fyne. Took one site from ~#20 to top 3 on TripAdvisor; 80% NPS on mystery-diner measures. The stakeholder management and operating-under-pressure foundation that engineering bootcamps don't teach.

Credentials

Certified & Continuing · always shipping

CompTIA A+ · Python (Codecademy, Google/Coursera) · Configuration & the Cloud · Year 1 BSc Biology, University of Portsmouth. Currently deepest in agent architectures, evaluation and LLM ops — learning by building, in production.

Transmission open

Let's put AI
to work.

If you need someone who ships production AI and gets a whole organisation using it, talk to me.

aaronhaydon1@me.com LinkedIn ↗