Highlights
- Deployed and maintained the system on-site in a secure, air-gapped mailroom with no remote access, working directly with non-technical staff and resolving in-place issues
- Built a Python/SQL system automating retrieval and aggregation of 800,000+ appointment letters from S3, ensuring compliance with a contractual 6-day mailing window and saving an estimated 50,000+ hours
Project Narrative
Every veteran with an upcoming DBQ appointment must be notified by a physical appointment letter, and the account carries a contractual obligation to mail that letter within six days of the appointment being scheduled. When a Performance Work Statement (PWS) update made this a firm requirement, meeting the six-day window at scale became a compliance necessity tied directly to the account’s revenue. Appointments that weren’t properly noticed put billable work at risk. The original process was entirely manual: mailroom staff downloaded each letter through the same web application used for all MDE work, one case at a time. As appointment volume grew, that approach couldn’t keep pace with the mailing deadline.
I wrote a Python script that queried the database to identify which appointments still needed letters, pulled the corresponding letters from S3, and prepared them for mailing. Rather than reproduce the one-at-a-time manual flow, I aggregated letters by veteran, so that a veteran with multiple upcoming appointments had all their letters downloaded together letting the mailroom consolidate mailings, optimize groupings, and cut the volume of physical mail sent.
The mailroom is a secure environment with no remote access and limited internet, so I couldn’t deploy or troubleshoot the way I normally would. I deployed the system on-site and in person, with minimal outside assistance, and worked directly with the non-technical mailroom staff who would operate it day to day. That included solving problems in place as they came up, such as a BitLocker issue that surfaced during deployment and had to be resolved on the secure machine itself.
By PY26 the system had processed and tracked 800,000+ appointment letters, and by streamlining this and related ancillary tasks across adjacent workflows it saved an estimated 50,000+ hours.