Oliver Wakeford

What I do.

Alongside the master's, I work as an AI engineer inside two companies and for a few freelance clients. Mostly that means building the systems a team works in every day, then helping the people in it get comfortable with them. At one of the companies it's security and pre-sales work instead.

The research is on the academic page, and what I've built is on projects.

Chiron Consulting

AI Engineer · Dec 2025 to present

Marketing consultancy · remote

  • Claude Code
  • Git
  • Python
  • BM25 retrieval

I built the shared workspace Chiron's content team writes in, and the retrieval system they draft with. The workspace is a private Git repository everyone opens in Claude Code. Each client brand has its own folder with its context and voice notes, and there are skills for the jobs the team does every week, like drafting posts or turning a meeting recording into notes. Opening a session pulls the latest version and closing it pushes, so a change one person makes reaches everyone within seconds.

Drafting starts by finding a few of the client's own published posts to use as examples, then Claude writes the draft and checks its own work. My M1 internship research tested that retrieval step offline and checked the automatic voice score against blind human ratings. The retrieval study has the full write-up.

I ran the team's onboarding in July and wrote the setup guides. The team runs it day to day now, and I still maintain it.

Acvium

AI Security & Governance Consultant · Mar 2026 to present

Cybersecurity consultancy, formerly ACE-NSI · part-time, remote

  • Python
  • Claude API
  • Zoho CRM
  • pytest

At Acvium I designed their method for assessing Shadow AI in the regulated firms they advise, which is staff pasting work into AI tools their company never approved. It lists the AI tools in use and rates each one on how sensitive the data going in is and whether any control covers it. That gives a risk level. The findings then map to GDPR and the EU AI Act, plus NIS2 and DORA where they apply. I wrote it up as a sample report on a made-up company.

I also tested two AI security platforms hands-on. One caught a test string typed into ChatGPT and warned the user as it happened, and the other listed the agents and MCP servers in a demo environment and graded the risk of each. From that I wrote a guide to which control fits which situation.

On the pre-sales side, I built a Python pipeline on the Claude API that researches target companies and keeps the CRM in step with the outreach.

Freelance

AI Engineer · May 2026 to present

Claude Code and AI builds · remote

  • Claude Code
  • Python
  • Claude API
  • MCP

For small businesses and individuals, I set up Claude Code around how they actually work. Claude Code reads their files with them while they explain how things run, and from that we agree on one working folder it reads every time, with skills for the jobs they repeat. The skills draft messages but never send them. After the setup it's usually coaching calls, where they do the work and I help.

I also build retrieval (RAG) systems over a company's own documents, ways to test whether an AI system's output is good enough, automations and small agents for repetitive work, and small AI web apps.

Before any of this, there was a massive run of student jobs: bar work, waitering, assistant teaching, au pairing, delivery driving, taxi services, gardening, and a lot of manual labour along the way. These taught me a thing or two, and I'm glad to be doing the AI work instead.

Tell me what you're working on.

We can have a chat about what you need.