Early-warning dashboards are now common in higher education, but many published systems either stop at a bare risk label or remain prototypes that never meet the practical demands of a real college: separate institutions, staff roles, secure sign-in and data-protection law. This paper presents ClassCircle, a web platform that grew out of the author’s earlier browser prototype and now serves several institutions from a single Rocky Linux virtual machine. Each institution has its own SQLite database behind one sign-in page, and the server alone decides what each of five roles may see. Risk is estimated with an L2-regularised logistic regression over seven academic and engagement indicators. Each 0–100 score is shown with its exact per-feature contributions and with rule-based next steps, and every summary number on a dashboard opens the list of students behind it. With no real student data available, the evaluation covers only measured results: 189 back-end tests, 85 browser checks, 25 acceptance scenarios and 34 attack checks passed; on a synthetic cohort of 120 students the model reached a mean AUC of 0.934 over 100 stratified splits, against 0.849 for the rules alone; and a burst of 120 simulated students writing at the same moment completed 360 requests without an error on a two-core test machine. The paper makes no claim that any student improved. Instead, it fixes in advance how improvement and possible harms will be measured in a staged pilot, and it maps the platform’s controls to India’s Digital Personal Data Protection framework.

Aqdus Shaikh Zahoorahmed, Dr. Manoj Singh, "ClassCircle: Design and Verification of a Multi-Institution Learning Analytics Platform with Explainable Early Warnings" International Journal of Scientific Research and Engineering Development, V9(5): Page(1416-1426) September - October 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.