$2.4M in DoD Contract Work, Off Paper and Into One System
ACME Manufacturing is a stage name — identifying details are changed to protect the operation. The client, the system, and the numbers are real.
The situation
Two people run ACME Manufacturing, a precision-parts contracting business serving the Department of Defense. They source, machine, and ship parts — brackets, assemblies, hardware — to the Defense Logistics Agency and other government customers. Every shipment ties back to a National Stock Number, a contract line item, and a government payment schedule. Contracts run through PIEE, the DoD's procurement system. Payments come through DFAS.
The owner handles production and shipping, and fills out a paper form for every job. The operations manager handles everything else: customer orders, subcontractor coordination, government inquiries, cash flow. Both know the domain cold. Neither had any interest in learning complicated software.
The whole operation ran on paper, spreadsheets, and email. Active jobs lived in Excel. Government inquiries got answered by hand. DFAS payments got reconciled against internal records manually, line by line — hours every week.
What was actually wrong
The system worked — but most everything stayed in the owner's head. No live view of which jobs were behind schedule. No list of subcontractors at risk. No warning when an inquiry was about to hit its deadline. No flag when a payment didn't land. When something fell through the cracks, it fell hard.
Off-the-shelf software wasn't the fix, either. Generic tools ask a two-person shop to operate like a fifty-person company: weeks of onboarding, permission tiers, screens full of fields nobody needs. They needed the opposite — software shaped around how they already worked.
What I built
ACME Smart Log: a purpose-built operations platform. One developer — me — from schema design to production deployment.
The home screen is a morning standup board. It surfaces what needs attention today: jobs that should have shipped but have no packing slip, inquiries due within 48 hours, payments stuck in "researching" past 30 days, subcontractors at risk, unmatched receipts, open discrepancies. The operations manager opens the app and knows where the day starts.
The paper form stayed. The owner photographs it, and Claude's vision API pulls the job number, customer order number, NSN, quantity, price, and ship date from the image. The data lands in seconds. Nobody re-keys anything.
Government paperwork gets the same treatment. The platform ingests three formats: a 48-column DFAS payment extract, DFAS statement PDFs, and subcontractor delivery receipts. Every upload previews exactly what will change before anything commits. No import silently overwrites good data.
A reconciliation engine joins payments, packing slips, and job orders that arrive from different sources at different times. It flags math errors, payment amounts that don't match shipped quantities, and shipments with no matching payment. The operations manager reviews each flag, resolves it, adds a note — and those notes survive every re-run.
Around that core: inquiry tracking with deadline countdowns and response drafts that auto-fill carrier, tracking number, quantity, and ship date from packing slips already on file. A dossier page for every job. Vendor invoice tracking. A directory of 162 suppliers, machine shops, carriers, and government contacts, each with full relationship history. Four reports with CSV export.
Login is a magic link — no passwords to forget, no rotation policy to babysit. Live sidebar counts update across browser tabs in under 100 milliseconds. Nobody hits refresh.
What changed
- About $2.4M in contract work tracked in one system
- 419 job orders under management
- Roughly 3,600 payment transactions reconciled against government records
- 20+ pages and views, built and maintained by a single developer
- Live in production since early 2026, handling daily operations
Mornings no longer start with digging through spreadsheets and email to find out what's at risk. The paper form still gets filled out on the production floor — it just stops being the only record seconds later. And every inquiry, payment, and shipment now leaves an audit trail.
What this says about how I work
I design around the real constraint — here, two expert users with zero patience for software ceremony. That constraint produced a cleaner system than most enterprise tools I've seen. And I learn the domain before I model it: the platform works because it matches how DoD contracting actually runs, not how a template assumes it does.
The domain also ran deeper than one client: the DLA knowledge this engagement built became the seed of RFQ Hunter, and this client is set to become one of its customers.
The paper form is still on the floor. It just goes somewhere now.