05 / FIELD NOTES

Ideas worth exploring.

01 / FIELD NOTE
Machine learningNOTE 01

Accuracy is a starting point, not a verdict.

Precision, recall, class imbalance, and the uncomfortable questions a single score can hide.

By Sarvan CRead article
02 / FIELD NOTE
ML researchNOTE 02

Learning to reverse the noise.

How a noise schedule, a denoising network, and repeated sampling combine to generate images.

By Sarvan CRead article
03 / FIELD NOTE
EsportsNOTE 03

The match is more than the scoreboard.

Reading momentum, objectives, and player choices without reducing a team to one statistic.

By Sarvan CRead article
04 / FIELD NOTE
Game developmentNOTE 04

Good game feel is engineered.

Feedback loops, input latency, readable states, and the small details that make controls feel right.

By Sarvan CRead article
05 / FIELD NOTE
Off the clockNOTE 05

Sometimes the best debug tool is a walk.

Why a deliberate break can help you return to a stubborn problem with a better question.

By Sarvan CRead article
SC.

Written to understand things more deeply. Updated as the ideas evolve.

More experiments on GitHub