From a hand-built certificate to a closed loop.

Certificates were created by hand. Every time a student passed an exam, a staff member built the certificate themselves. At thousands of passes a year, that was a full-time drain with built-in delay and human error.

The data lived everywhere and nowhere. Assessment results sat in one system, coursework in another, contacts and outreach in a third, and tracking spreadsheets in between. Nobody could answer a question as simple as how many students were certified in one state last semester without manual reconciliation.

Coordinators and teachers had no visibility of their own. The people running the program in each state, and the teachers running the classrooms, had no way to see results, download certificates, or know when a certification was about to lapse.

Scale had outrun the process. The program spans hundreds of schools across more than a dozen states today. The manual process was not just slow, the organization had never had this data at real accuracy, at any scale.

The pipeline above starts the moment a student sits an exam. It said nothing about how the student got into the course in the first place, and that turned out to be where the data quality was actually being lost.

Students used to type their school and teacher into a free-text box, and free text produces exactly what free text always produces: the same school spelled three different ways, with nothing downstream able to tell them apart. Since the system of record stores a student's teacher as a real link rather than a typed name, a mistyped instructor could never be connected to the right record at all.

The fix was a personal signup link for every teacher, one link, reusable every year, with the teacher and school already filled in because the link identifies them. A student opening it sees only the courses that teacher is actually certified to deliver, nothing else.

A nightly safety sweep still checks for anyone who found a side door in anyway, and proposes a match against the real teacher roster rather than guessing. Every proposed match is reviewed and approved by a person before anything is written back. The sweep itself only ever reads. An automated system confidently reassigning a student between teachers on a fuzzy name match would do more damage than the problem it was built to solve.