"Ecosystem engineer" is an ecology term, and it's meant to be descriptive not precisely literal. It doesn't necessarily indicate any intention. TFA did a poor job of conveying the fact that this is a field-specific usage, not a description of "engineering" by animals.
Some animals have disproportionately large effects on the integrity of their ecosystem - disproportionate to their biomass and physical presence, at least. These animals are called "keystone species". Apex predators are often keystone species due to their effects on prey behavior and their strong actions as selective pressure.
Some keystone species provide specific metabolites that are critical to their ecosystem. You could argue that the organisms that allow termites to digest cellulose are probably keystone species. Nitrogen-fixing organisms would be there as well. Those examples, though have very localized effects.
Some species are keystones for reasons other than simple predator-prey relationships. Animals who significantly physically change their environment are frequently referred to as "ecosystem engineers". Burrowing animals whose dens are required by other critters are one example (in the US, tortoises and ground squirrels are notorious for this). Beavers, as mentioned above, are as well. Underwater, the composition of the water itself is the environment, and changing that composition can have a huge effect on the ecosystems involved. Use of the term "ecosystem engineer" in this context is simply meant to convey how critical whales are to maintaining a healthy and diverse ocean ecosystem, despite previous assumptions that their relatively low biomass (because of their rarity) implied that they were not particularly integral.