If the Maritime industry is to adopt carbon credits and carbon credits trading (the market for which our startup was ultimately aiming at), it needed a way to systematically track emissions data for all vessels. Ideally, this would occur in a manner that's frictionless (permissionless) yet reliable (as self-reporting only goes so far). Hence, the goal of predicting vessel emissions from "publicly available" satellite data.

I built a model to estimate CO2 output from AIS satellite tracks and vessel specification sheets (engine types, capactiies, etc.). It's physics combined with data modelling.

The end goal of the startup was to be the marketplace for carbon credits. But we couldn't offset what isn't accounted for, and so this was the accounting layer.