Engineering principles

AI moves fast.
Experience keeps it pointed
in the right direction.

I combine 15+ years of building for the web with AI-native development. The tools make the work faster; judgment, architecture and accountability remain human.

01

Position

AI is part of the workshop. Experience decides what belongs in the product.

I do not measure engineering by how many lines were typed by hand. I measure it by whether the result is useful, understandable, resilient and ready to own.

AI gives me more ways to explore, prototype and verify. Years of experience tell me which questions to ask, which shortcuts are dangerous, and when the simplest solution is the strongest one.

Experience
15+ years
Method
AI-assisted
Direction
Human-led
Responsibility
Always mine
02

Principles

The standards I use to turn an idea into software that can keep earning trust.

  1. 01

    Understand before automating

    The first job is to clarify the real problem, the user and the constraint. A fast solution to the wrong problem is still the wrong solution.

  2. 02

    Let architecture earn its complexity

    Every abstraction has a cost. I start with the smallest sound system and add structure when the product—not fashion—requires it.

  3. 03

    Prefer clear code over clever code

    Names should explain intent, boundaries should be visible and future changes should be unsurprising. Maintainability is a feature.

  4. 04

    Build security into the shape

    Validate inputs, minimise access, protect data and design failure paths early. Security is not a final checklist applied to an unsafe foundation.

  5. 05

    Test the behaviour that matters

    Tests should protect critical journeys and risky boundaries, not merely inflate a coverage number. Automation supports judgment; it does not replace it.

  6. 06

    Ship something that can be operated

    Useful logs, predictable deployments, sensible defaults and concise documentation are part of the product—not work left for later.

03

AI Practice

Transparent by design: what the tools accelerate, and what I never delegate.

Used deliberately, reviewed critically.

AI is a powerful collaborator.
It is not the accountable engineer.

AI accelerates
  • Exploration and prototypes
  • Implementation drafts
  • Refactoring options
  • Test-case generation
  • Documentation and review
I own
  • Product decisions
  • System architecture
  • Security and privacy
  • Technical trade-offs
  • The code that ships
04

Process

A short feedback loop keeps momentum high without trading away control.

  1. 01

    Frame

    Define the problem, constraints and measure of success.

  2. 02

    Shape

    Choose the simplest architecture that leaves room to grow.

  3. 03

    Build

    Work in small, reviewable increments with AI in the loop.

  4. 04

    Prove

    Test critical paths, edge cases, security and real behaviour.

  5. 05

    Ship

    Deploy predictably, observe the result and improve from evidence.

Move fast where speed is reversible. Slow down where trust is at stake.

— The rule behind the workflow