Abstract. Orphaned accounts, where people who have left still hold active access, are among the most common and preventable security gaps in higher education. This paper describes how a mid-sized community college reduced former employees with active accounts from about 670 to zero. It replaced manual, ticket-driven offboarding with an event-driven lifecycle engine fed by the statewide PeopleSoft HCM system. The central lesson: the code was the easy part; the timing was the design.
Keywords: identity lifecycle, joiner-mover-leaver, PeopleSoft, change data capture, access governance
1. The problem
Accounts were created and removed by hand, triggered by emails and tickets from HR. Every hand-off was a chance for a step to be missed. Over years, the gap between “left the college” and “lost access” added up to hundreds of live accounts belonging to people who no longer worked there.
2. The approach: snapshot, diff, act
The engine follows a three-step pattern:
- Snapshot. Take a full picture of the employee population from the HR system of record.
- Diff. On each run, compare the new picture to the last one and classify every change: new hire, rehire, job change, separation, name change.
- Act. Hand each change to a task runner that creates, updates, disables or reactivates the account in the directory.
Because the engine reacts to changes rather than re-processing everyone, it stays fast and its actions are easy to audit.
3. Timing is the real design problem
HR records are often entered before a start date, edited after it, or back-dated. Act too early and accounts exist for people who never start. Act too late and new hires sit without access on day one. Working sessions with HR, using an interactive timing explorer, set when each step fires relative to the effective date and the date the record was entered. [Add the final timing rules.]
4. Edge cases teach the system
Every new employee type exposed a gap: volunteers, trustees, external instructors, staff who are also students. The engine surfaces records with missing fields instead of failing silently, so each edge case becomes a rule rather than an outage.
5. Results
- Former employees with active accounts: about 670 → 0
- Onboarding, deactivation and reactivation automated end to end
- Preferred names flow from HR into accounts automatically
6. Lessons
Treat HR as the system of record and the lifecycle engine as its listener. Design timing with the people who own the data. Make every exception visible.