Engineering leadership for trustworthy AI-assisted delivery

I build the systems that make AI-assisted software delivery trustworthy, measurable, and scalable.

My work combines engineering leadership, agent orchestration, policy-as-code, context engineering, CI/CD governance, software supply-chain security, evaluation, and delivery intelligence. I help organisations adopt AI without surrendering quality, accountability, or human judgement.

What I Build

Governed AI-assisted delivery

GitHub-native PR governance, runtime-independent review, human approval gates, risk classification, evidence bundles, and release-readiness workflows for teams delivering software with AI agents.

Executable engineering governance

Policies and standards turned into machine-readable contracts, JSON Schemas, conformance tests, CI/CD quality gates, security checks, and audit-ready change records.

Agent evaluation and context quality

Benchmark harnesses and framework adapters that test agents against real software engineering tasks, measuring specification conformance, evidence quality, regression risk, first-pass success, and the need for human rework.

Engineering leadership and delivery performance

Team scaling, technical leadership, roadmap predictability, DORA and SPACE metrics, flow optimisation, release governance, reliability engineering, and continuous improvement.

Platform engineering and developer experience

Internal developer platforms, developer portals, service catalogues, golden paths, self-service workflows, CI/CD automation, and the practical reduction of friction that slows teams down.

Current Work

I am building a connected set of systems for governed agent-assisted delivery:

• A GitHub-native PR governance platform that produces policy-tagged, audit-ready evidence packs.
• A machine-readable specification for governed software change records, controls, approvals, exceptions, and evaluation results.
• Pathfinder AFLL, a framework-neutral benchmark lab for agent and multi-agent systems.
• A shared-agent-skills platform for reusable skills, validation, evaluation, conformance, trusted publishing, and governed dispatch.

Together, these projects explore how engineering organisations can move from AI experimentation to an operating model with clear permissions, measurable quality, inspectable evidence, and human control where risk demands it.

Featured Work

These are the most useful places to see the approach in practice:

• Pathfinder AFLL - a framework-neutral benchmark lab for evaluating agent and multi-agent systems against governed software engineering tasks.
• Governed PR Evidence Pack - a structured pattern for turning AI-assisted pull requests into auditable release decisions.
• Context Quality Benchmark Harness - a benchmark pattern for measuring whether agents receive faithful, relevant, useful context.
• Engineering Health Radar - delivery, quality, risk, flow, and governance signals for engineering leaders.
• Multi-Agent SDLC Control Plane - a reference architecture for governed agent-assisted delivery across planning, implementation, review, evidence, and release readiness.
• Role Fit Explorer - an evidence map for helping hiring managers connect role needs to relevant experience, projects, and interview questions.

Work With Me

I bring the combination of engineering management, hands-on technical direction, platform thinking, and delivery governance needed to turn ambitious AI initiatives into systems teams can operate with confidence.

Explore the projects, read about my background, or get in touch.