Engineering notes

Decisions, trade-offs and lessons from shipped products.

Clear, evidence-led case studies from Mapify, Henri, AxIT and Aimi—how a vague requirement became a concrete problem, which trade-offs shaped the architecture, where AI accelerated the work, what changed in production and what I would carry into the next system.

System designProblem framingArchitectural trade-offsAI-assisted engineeringCross-platform systemsEvidence-led delivery

All case studies

11 articles · newest first

Mapify · Henri · AxITDecember 2025–November 2026

Engineering vibe coding: AI inside guardrails

How I moved from Cursor pairing to Claude Code and Codex agent teams by treating AI as a capable engineering team member: explicit behaviour, architecture, design rules, reusable workflows and evidence—not prompts alone.

18 min readRead case study
AxITOctober–November 2026

Two-phase QA for metric algorithms and app behaviour

How I designed a two-phase quality process: AxLib focuses scarce domain-expert time on production-sample evidence and version differences, then SBN Testing lets app testers validate exact-build behaviour without rechecking the maths. Developers fix reported bugs while that testing continues, instead of waiting for a finished pass to be handed over.

11 min readRead case study
Mapify · Henri · AxITJanuary–September 2026

AI debugging: from plausible patches to reproducible evidence

How I evolved AI-assisted debugging from code inspection and likely fixes into a reproducible workflow: exact bug data, real product paths, layered observability, regression tests, computer-use verification and physical-device evidence where the boundary requires it.

17 min readRead case study