<About>

I build and own production AI and automation systems that remove the operational overhead people have come to accept as normal. At Credit Karma that work saved $500k+ and 2,500 hours of manual effort a year, and cut IT ticket volume 46% year on year.

I specialise in AI automation and internal AI enablement: agentic workflows, enterprise automation architecture with Workato and n8n, and rolling AI out safely to non-technical teams. I'm looking for a senior AI automation or AI enablement role in London, on-site or hybrid.

Coding agents like Claude Code write the code, and I own the engineering around it. That means a plan, hooks that limit what agents can touch, and pull requests reviewed by me and CodeRabbit. Automated tests run in GitHub Actions and builds are containerised through GitHub, with a sandbox before production. Permissions are least-privilege and secrets stay out of code. Once something is live it has monitoring, audit logs and documentation, and every release can be rolled back.

Off the clock I'm a music producer and DJ, a gym regular, and an obsessive traveller and foodie.

<Experience>

Credit Karma

  1. Jul 2019 — Aug 2026

    Senior AI Automation Engineer

    • Generated $500k+ in annual cost savings and eliminated 2,500 hours of manual work per year through cross-departmental automation programmes built on Workato at enterprise scale, working with Finance, Legal, HR, Product and Engineering.
    • Built a Slack-based AI support agent used across the whole company. It answered with RAG on Google Cloud (Vertex AI Vector Search, Gemini) and acted through its own MCP servers for Okta, Jamf, Jira and Google Workspace to complete multi-step requests end to end. It handled 45,000 tickets, cut median resolution from two days to under two hours, reduced IT ticket volume 46% year on year and reclaimed 3+ FTEs of capacity.
    • Designed a human-in-the-loop feedback loop so no question needed a human twice. Unanswered tickets went to a person; once the requester confirmed the fix and it was reviewed, the resolution was written back to the vector index within minutes.
    • Opened the same MCP servers up company-wide through an internal MCP portal, version-controlled in GitHub. Anyone could use them from Claude, ChatGPT or Gemini, limited to their own permissions.
    • Ran all-hands sessions and presented at product showcases so non-technical teams knew how to use the tools.
    • Automated temporary system access with Security, replacing a manual, approval-heavy queue with self-service: 6,238 requests, median resolution cut from 3.7 hours to 3 minutes and $240k a year in overhead recovered. Every grant was logged and reviewed for external audit.
    • Built a weekly AI-written digest of IT support trends from BigQuery. Leaders read it in their inbox without opening a tool, and readership grew to the SVP of Engineering.
    • Engineered a data pipeline with the BI team to extract and analyse company-wide calendar data, which showed senior leadership thousands of hours lost to inefficient recurring meetings.
    • Workato
    • GCP
    • Vertex AI
    • Gemini
    • RAG
    • MCP
    • BigQuery

  2. Senior IT Support Engineer II

    • Architected the company's first AI-driven enterprise search platform, enabling self-service knowledge access across all departments.
    • Served as company-wide SME for Workato and led a zero-downtime migration to Okta Identity Engine across Engineering, Security and IT.
    • Workato
    • Okta

  3. Senior IT Support Engineer

    • Designed and managed a Slack-based IT support chatbot handling 7,500+ requests monthly. It evolved into the RAG-powered support agent, then into agents that act through MCP.
    • Built Workato automation across Okta, Slack, Jira and Google Workspace to connect business-critical systems and eliminate manual handoffs.
    • Workato
    • Slack
    • Okta
    • Jira
    • Google Workspace

  4. IT Support Engineer II

    • Owned multi-site IT infrastructure projects from design through implementation, and built core processes across user lifecycle, security and automation.

Earlier

Refinery29

  1. May 2017 — Jul 2019

    Lead IT Support & Infrastructure Engineer

    Led international IT support and infrastructure across London, Berlin, New York, LA, and Toronto, progressing from International Support Engineer to lead.

Dropit Shopping

  1. Aug 2016 — Apr 2017

    IT Manager

    Owned IT operations and systems for a fast-growing retail technology startup.

Apple

  1. Oct 2014 — Aug 2016

    Genius / Technical Specialist

    Diagnosed and resolved hardware and software issues and delivered technical support in a high-volume retail environment.

<Projects>

Selected Work

  • Slack · Vertex AI · Gemini · MCP

    Slack AI Support Agent

    A Slack-based AI support agent used across the whole company. It answered with RAG on Google Cloud (Vertex AI Vector Search, Gemini) and acted through its own MCP servers for Okta, Jamf, Jira and Google Workspace to complete multi-step requests end to end. It handled 45,000 tickets, cut median resolution from two days to under two hours, reduced IT ticket volume 46% year on year and reclaimed 3+ FTEs of capacity. Unanswered tickets went to a person. Once the requester confirmed the fix and it was reviewed, the resolution was written back to the vector index within minutes, so no question needed a human twice.

  • MCP · GitHub · Okta · Jamf · Jira

    Internal MCP Portal

    An internal MCP portal, version-controlled in GitHub, that opened the support agent's MCP servers for Okta, Jamf, Jira and Google Workspace up to the whole company. Anyone could use them from Claude, ChatGPT or Gemini, limited to their own permissions.

  • Workato · REST APIs · Okta

    Temporary-Access Automation

    Self-service temporary system access, built with Security to replace a manual, approval-heavy queue. It handled 6,238 requests, cut median resolution from 3.7 hours to 3 minutes (74× faster) and recovered $240k a year in overhead. Every grant was logged and reviewed for external audit.

  • BigQuery · Gemini · Slack

    Weekly AI Digest for Leadership

    A weekly AI-written digest of IT support trends from BigQuery. Leaders read it in their inbox without opening a tool, and readership grew to the SVP of Engineering.

Built Independently

  • MCP · Cloud Run · GCS · n8n

    Personal AI Memory System

    A RAG system that gives Claude persistent context across conversations. An Obsidian knowledge base on Google Cloud Storage, served through a remote MCP server on Cloud Run, with scheduled maintenance agents running on a headless Mac Mini and n8n. Versioned, authenticated, and in daily use.

  • MCP · Cloud Run · Cloud SQL

    Workout & Body-Composition MCP

    A self-hosted training platform: an MCP server on Cloud Run backed by Cloud SQL (Postgres), with a progressive web app served from the same service. It uses token-based auth, keeps no server state and scales to zero when idle. It logs every session and InBody scan, and reports progress against goals and recovery patterns.

<Technologies>

  • Vertex AI
  • Gemini
  • Claude API
  • RAG
  • MCP
  • Claude Code
  • CodeRabbit
  • GitHub Actions
  • GCP
  • Cloud Run
  • BigQuery
  • Postgres
  • Workato
  • n8n
  • REST APIs
  • Okta
  • Jamf
  • Slack
  • Jira
  • Google Workspace

<Certifications>