Hands-on Chief AI Officer · Freelance

Harness engineering expert

Harness engineering for your team.

I am most at home working on harnesses: designing agent environments, managing context, building tools and writing code. I configure these systems and work with developers to use them in production.

My experience as CTO and VP Engineering also helps me work with leaders on AI adoption, team organisation and training. I can work with you on a specific technical problem or take on a wider, hands-on Chief AI Officer role.

I build and contribute to open-source tools, with daily production use of generative and agentic AI since July 2025.

01 / YOUR SITUATION

Where I can help

01 / Build your harness and agentic software factory

I have built several agent harnesses for TypeScript and Python stacks, in monorepos and multi-repo environments of varying complexity. The work covers context, instructions, tools, skills, hooks, permissions and checks, including Claude Code and Codex configuration.

I can help design and build the complete agentic software factory: shared company knowledge, task preparation, agent orchestration, development environments, tests, review, delivery and observability. We define how those parts work together, who maintains them and which decisions need human approval.

Context & knowledge
Tools & permissions
Checks & review

02 / Technical audits & AI readiness

I carry out technical audits and AI readiness assessments: codebases, architecture, security, agent configuration and team practices. I identify what is ready for AI-assisted work, what needs strengthening and which changes to prioritise. I share the findings and remediation priorities with engineers and leadership.

03 / Improve delivery speed and maintain quality

I work alongside developers on real deliverables, from configuring the agent environment to writing code and delivering changes in production myself. Together, we work through code reviews and corrections so the developers can use and maintain the workflow themselves.

04 / Decide where to start

Identify useful AI applications, assess your current practices and choose what to tackle first. We turn that assessment into a prioritised roadmap with costs, responsibilities and criteria for evaluating the work.

05 / Put governance into daily practice

Define what agents can access and do, who approves sensitive actions, and who owns the resulting work. Connect those rules to budgets, team responsibilities and the way systems run.

Finance, HR and security join the work where their decisions and practices are affected.

06 / Evolve your organisation, roles and career paths

As agents change how work is done, I help you rethink team responsibilities, collaboration and decision-making. With leadership, tech leads and HR, we define what each role owns and how people develop as those roles evolve.

Together, we build skills frameworks, career paths and training plans. We define what people need to learn for their roles, how they will practise and how to assess their progress.

Decide: priorities and responsibilities. Build: context, tools and checks. Adopt: training and shared practices. Review and adjust feeds back into decisions. — Priorities, engineering and team practices inform each other through review.

02 / TRAINING

Advanced training in harnesses and AI-assisted development

My training speciality is advanced AI-assisted engineering: building harnesses, managing context, reviewing generated code and controlling agent workflows.

Practice on real work

Workshops combine explanations with practical exercises: investigate a defect, test a correction, configure permissions, check a hook's behaviour, or examine the cost of an accepted change.

From executives to developers

I have delivered training to executive committees, developers, lead developers and senior engineers, adapting the content and technical depth to each audience. Leaders learn to assess opportunities and steer adoption; engineers practise configuring, using and checking agent workflows. I also design programmes for product and non-technical teams.

Build a training plan

The training plan addresses skill gaps and changing responsibilities, with learning objectives by role, prerequisites, practical exercises, mentoring and checkpoints. This can support new career paths as well as immediate delivery needs.

Choose the format

Bootcamps, workshops and mentoring can stand alone or support a wider adoption engagement, in-house or in open sessions.

What we work on by role

Developers

Build, test and review code with agents, and learn to maintain the workflow.

Tech leads

Design the architecture, clarify responsibilities as roles change and establish shared engineering practices.

Executive teams

Understand AI, choose where to invest and assign responsibilities for adoption.

03 / ENGAGEMENT FORMATS

How we can work together

A scoped technical assignment

An audit, a roadmap, a harness setup or a defined technical problem, with an agreed scope and handover.

Working alongside your team

Implementation, coaching and mentoring on real work, with documentation and practices the team can continue using.

Fractional Chief AI Officer

As a fractional Chief AI Officer, I connect investment decisions, technical delivery, governance and organisational change. I support leadership, role and career development, and the training plan. We agree which decisions I own and which remain with your leadership team.

Training

A standalone programme or a series of sessions connected to your team's adoption work.

We can work out the scope and format together. You do not need a finished brief.

04 / IN PRACTICE

Examples from my work

~60 repositories audited

Technical audits & AI readiness

I carry out technical audits and AI readiness assessments. One technical and security audit covered around 60 repositories for a SaaS company, using a 12-axis maturity framework, with a remediation plan presented to leadership.

Executive teams → senior engineers

Training across roles

I have trained executive committees, developers, lead developers and senior engineers. One example: a Claude Code bootcamp for 30 people across 5 squads at a professional training company.

TypeScript · Python

Harnesses across stacks

I have built several harnesses for TypeScript and Python codebases, spanning monorepos and multi-repo environments. The configuration adapts to the complexity of the codebase, its tools and the way the team delivers software.

Hands-on delivery

Configuration, delivery & adoption

My support includes configuring agent environments and delivering production changes myself alongside teams. In one ongoing engagement, I trained teams, configured their harness, created a dedicated repository and built a shared company knowledge base.

Tell me what you are trying to change

Describe what you are building or trying to change. A harness, an agentic factory, a training plan, or wider questions about AI direction, organisation and careers: we can start from your situation.

Let's discuss your situation →