Attendance management is a critical operational process in educational institutions; however, conventional manual registers and loosely controlled digital systems often suffer from proxy attendance, delayed updates, and lack of institutional control. This paper presents Markbase, a staff-controlled, AI-assisted, web-based attendance management system designed to replicate real-world academic workflows through a timetable-driven operational model. In the proposed system, attendance sessions are dynamically activated based on division-specific academic timetables and can be initiated only by authorized staff members, ensuring strict procedural governance. The system integrates artificial intelligence–based facial recognition for secure student authentication during login and attendance marking, significantly reducing identity fraud and proxy attempts. A rule-based attendance engine automatically assigns Present, Late, or Absent status using time-bound session validation, eliminating manual manipulation. Markbase implements role-based access control supporting Administrator, Staff, Student, and Parent roles, each provided with role-specific dashboards, analytics, and reporting interfaces. The backend is implemented using a Python-based RESTful architecture with SQLite database support, while the frontend utilizes ReactJS to provide a responsive, browser-based interface suitable for both desktop and mobile devices. The proposed architecture ensures scalability, reliability, and transparency while maintaining administrative authority over attendance processes. The system is designed for realistic institutional deployment and serves as a comprehensive final-year engineering or diploma-level academic project demonstrating the integration of artificial intelligence, web technologies, and automated academic workflow management.