Document Digitization
NASA JPL . Chevron Inc.

The effort aimed at converting operator (of large machinery) reports into clean data within a well structured database. Such reports are usually heavy on jargon and written in short hand, making it a nightmare for regular NLP (Natural Language Processing) systems. If solved, this had several applications within JPL and Chevron. We chose Chevron oil fields operator report digitization as the test case (and the test data). Once this data is well structured into databases it has huge financial implications (for strategic decision making and such..). Some interesting background; this use case has been worked upon at Chevron for 8 years (with multiple collaborations) without any real success when we took this task. Two years later we deployed a solution that was later used to digitize hundreds of thousands of reports. It was jointly patented by Caltech and Chevron.

​

I was the lead inventor for this product, jointly patented by Caltech and Chevron: https://patents.google.com/patent/US11790170B2/en