Methodology
Methodology
Embedded Files
AI-Assisted Engineering in Practice
AI-Assisted Engineering in Practice
The projects documented on this site demonstrate an AI-assisted engineering approach combining domain expertise, independent technical problem-solving, software engineering principles, and iterative collaboration with large language models.
I define the problems, develop the solution approach, make technical and business-logic decisions, and validate the resulting implementations. AI accelerates research, implementation, debugging, documentation, and testing, while human judgement remains central to the process. The objective is never to remove human oversight, but to increase engineering capacity while maintaining quality, traceability, and control.
AI supports:
research and analysis
prototyping
code generation
debugging
documentation
testing
All outputs are reviewed, validated, and refined before use.
Human-Led Decision Making
Human-Led Decision Making
AI can suggest approaches and accelerate implementation, but I retain responsibility for the engineering decisions that determine whether a solution is appropriate and safe to use.
I make the final decisions on:
architecture and solution design
business logic and transformation rules
risk assessment and safeguards
validation and testing
readiness for operational use
AI provides assistance while I remain accountable for the resulting solution.
Safety-First Automation
Safety-First Automation
Automation should reduce risk, not introduce it.
The development workflow incorporates mandatory safeguards for generated scripts:
✅ validation before execution
✅ transactional processing and rollback protection where applicatble
✅ audit logging and traceability
✅ version control
✅ human review and controlled execution
These controls are built into the development process rather than left to individual scripts to implement independently. The objective is to make automation reliable, reviewable, and controlled by design.
Iterative Development
Iterative Development
Projects typically follow an iterative development model:
Discovery: understanding the problem space, existing workflows, risks, and constraints.
Prototype: building an initial implementation to validate concepts and identify unknowns.
Validation: testing against real-world conditions while identifying edge cases and operational requirements.
Refinement: improving architecture, maintainability, usability, and resilience.
Operationalisation: adding auditability, governance, documentation, and repeatable deployment pathways.
This cycle repeats continuously as tooling matures.
Several projects began as targeted solutions to specific operational problems and evolved into reusable tooling or repeatable workflows as further requirements emerged.
Engineering with Curiosity
Engineering with Curiosity
Many of the most valuable projects documented here began as questions rather than formal requirements.
Examples include:
reverse engineering undocumented behaviours
exploring complex pricing models
identifying process bottlenecks
analysing configuration inconsistencies
designing automation around operational pain points
AI has become a powerful tool in enabling that exploration, helping transform curiosity into practical, measurable outcomes.
Lessons Learned
Lessons Learned
AI accelerates implementation: AI can dramatically reduce development time when problems are clearly defined.
Human judgement remains essential: domain expertise, risk assessment, and validation cannot be delegated.
Safety enables scale: the larger an automation ecosystem becomes, the more important governance and auditability become.
Curiosity drives innovation: many of the most successful projects started as experiments, investigations, or attempts to understand complex systems more deeply.
Philosophy
Philosophy
Use AI to accelerate learning, exploration, and engineering.
Use engineering discipline to govern AI.
Use automation to remove repetitive work.
Keep people responsible for the decisions that matter.
Page updated
Google Sites
Report abuse