Contamination report · Photo log
1,276 site photos from a CDBG-MIT warehouse project became a 32-entry, 83-photo log with 19 numbered findings during a two-day pilot, every draft saved as a revision. A coverage check over the photos left out surfaced three features the first pass had missed.
Result: Nearly three times the coverage of a typical approved report (about 12 entries and 36 photos)
Explosives survey · HUD ASD
153 tanks recorded across 7 projects; 139 run through HUD’s official calculator automatically with the PDF filed. A rule that also measures to the Off-Site corridor caught a tank 180 ft from it against a 238 ft separation distance — and every project with tanks was recomputed.
Result: 139 official ASD calculations · one compliance error caught and corrected
Site geometry · CAD to map
A project drawing ignored an 8.5° rotation of its own coordinate grid, placing the site about 8.5 m off. Refitted to the drawing’s grid and checked against LiDAR elevations at 2,269 survey spots, it now lands within about half a meter.
Result: Boundary error cut from 8.5 m to 0.53 m (median)
Linear utility · Erosion-control report
The design of a linear project arrived as CAD exports. The import placed every structure on the map and flagged the ones drawn away from their scheduled position; they went back to the designer as questions, not as silent corrections.
Result: Same-day report with 0 items left to confirm
Agency review · Comment rounds
346 reviewer comments extracted from 17 review PDFs totaling 9,839 pages, each with its reply thread and an “addressed” checklist filed on the project, so the team sees how many are done.
Result: Every comment accounted for, round after round
Karst · LiDAR screening
A 1-meter terrain model of a 241-acre site in Puerto Rico’s northern karst revealed 295 closed depressions — 35 inside the site, the deepest of them 5.9 m — and showed that only 14 of 48 mapped ponds actually sit in a sinkhole.
Result: 295 depressions mapped and classified in 80 seconds of processing