Blog

Thoughts & Stories

Sharing my experiences, learnings, and adventures in tech and beyond.

Browse by Topic

DevOps & Cloud Infrastructure (138)AI & LLMs (102)Web Development (81)ERP Systems (72)Backend & APIs (60)Software Development (52)Security (28)Career Reflections (27)Cloud (15)Product & SaaS (12)AI / Machine Learning (5)University (1)Hackathons (1)

Reviewing AI-Written Postgres Migrations Before They Run

Backend

September 202613 min read

Reviewing AI-Written Postgres Migrations Before They Run

An agent writes valid SQL that passes on an empty table and locks a live one. The lock every Postgres migration statement takes, and the review that catches it.

AI Pair Programming and the Junior Developer Skill Gap

Career

September 202611 min read

AI Pair Programming and the Junior Developer Skill Gap

The skill AI pair programming puts at risk is not syntax, it is the debugging loop. The five sub-skills that decay, a predict-first drill that preserves them, and what review has to ask.

September 202611 min read

Deterministic Hooks vs Agent Judgement in Claude Code

A hook fires whether the agent agrees or not, while an instruction only competes for attention. The five axes I use to decide which layer carries a rule.

Spec-Driven Development for ERP Requirements With AI Agents

ERP

September 202612 min read

Spec-Driven Development for ERP Requirements With AI Agents

ERP requirements arrive as edge cases, not specs. How I turn them into testable acceptance criteria, a worked-example table and an ambiguity log.

AI-Assisted Legacy ERP Refactor With No Written Spec

ERP

September 202613 min read

AI-Assisted Legacy ERP Refactor With No Written Spec

The module was nine hundred lines and its only specification was a commit message from 2017. Here is the method that survives that: characterisation tests first, then seams, then narrow agent-led extractions, and the four modules an agent never touches alone.

September 202612 min read

AI Code Review as a CI Merge Gate: What Should Block

An AI reviewer takes an afternoon to add to CI and months to calibrate, and the calibration is the whole problem. Here is the split between blocking and advisory findings, the shadow-mode measurement that earns a class the right to block, and the rule that takes it away again.

Showing 31–36 of 594 posts

Explore More

Projects

See the projects I've built

Experience

My professional career

Certifications

Professional certifications