The first perfect scorecard AT&T ever gave a turf vendor.

02The constraint

• ×10,000+ concurrent tower builds, each carrying hundreds of construction milestones — zoning, A&E, permitting, design, construction, test — across a dozen build types.
• ×AT&T's systems had no APIs and no SDKs. NORAD, CASPR, Treev, Business Objects. All of them closed.
• ×A dozen warehouses holding millions of dollars of antennas and equipment stacked to the ceiling, and no logistics system at all. Stock arrived, people took what they needed, projects went late.
• ×The Quickbase API could not define the variables I needed, and there was no time to build browser orchestration first. Version one ran on six laptops executing Excel macros in parallel.

03What I built

1. 01Headless browser sessions querying AT&T's API-less systems continuously and ETL'ing the results into one model, joined against projects: financials, milestones, NORAD scoping, BOMs, customer approvals, backup documentation.
2. 02A tracker grid with three date layers — plan, set by AT&T; forecast, set by the project manager; actual, set at closeout with the documentation attached.
3. 03A nightly routine that pushed at-risk unactualized forecasts out 48 hours and told the responsible project manager their milestone was in jeopardy.
4. 04A Motorola laser barcode system in C#, talking straight to Quickbase — receives, pallet moves, builds and breaks, location enumeration, project attribution — deployed across the warehouse network.
5. 05The eight-person engineering department around all of it, hired and led.

04Elapsed

Started

1 January 2012

Left

July 2013

Last counted on other people’s profiles

August 2026

05What it produced

The first turf vendor in AT&T's history to receive a perfect scorecard for timeliness. It won WesTower hundreds of millions of dollars in additional business.

The mechanism is the part worth telling, because it sounds backwards. Automatically pushing a date out is the sort of thing a client should hate. The opposite happened: WesTower's data became the only data in the program anyone trusted. When a competitor missed a date, AT&T could not tell whether the work was late or the number was stale. When we moved a date, it meant the work was late and someone was already on it.

In WIP meetings AT&T's own project managers would insist our numbers were wrong. They were newer.

Bechtel, Black & Veatch and General Dynamics had nothing comparable. Neither did AT&T.

In August 2026, fifteen people at seven companies listed Wesbase as a skill on their own profiles: WesTower, MasTec Network Solutions, Crown Castle, SureSite, Jacobs Telecommunications, T-Mobile, Elite ICT. Fifteen is a lower bound — it is what one Google search returns, and Google indexes only part of LinkedIn. When MasTec acquired WesTower, the teams kept it running alongside their own enterprise stack rather than migrating off it. I left in 2013.

I didn't put any of that there.