Notes

A defence of boring, idempotent ELT — why running the same SQL again on Monday morning should produce the same numbers, and what that costs to enforce in a real warehouse.

What two years of automation work taught me before I moved to a data role — which lessons were portable, which were a trap, and what I'd tell a junior RPA developer eyeing the same move.

Field notes from the TÜBİTAK-funded segmentation project. Less about architectures, more about the unglamorous decisions — normalisation, patch sampling, honest evaluation — that decide whether a Dice score means anything.