Biomimicry, not invention. He reasoned that the DNA is a vast library for transforming molecules from one to another, and therefore, ecosystems are capable of breaking down pollutants.
In practice, there are multiple vats. The first stage has algae growing, which sequesters the heavy metals. The next stages follow other kinds of ecosystems, such as organisms from swamps. He will mix samples from multiple ecosystems that normally don’t mix so that some kind of novel, self-organizing ecosystem can form around the pollutant.
Then it is measuring and monitoring the contaminants. With the superfund site, he was tracking presence of the top ten pollutants on the EPA list. However, he also shows how people can use much simpler, non-industrial tests — using samples from say, uncontaminated lake water nearby and use a microscope to see if the water being treated will kill those microorganisms. This allows for remediation to be executed by people who don’t have access to labs, but still need a way to test their water.
A much simpler version of this that follows the same design principles is capable of local, onsite treatment of ordinary black water.
Likely not what you’re describing, but this hypothetical is what came to mind.
Consider if organism A consumes organism B in a symbiotic way. E.g., organism B makes berries.
Then, we engineer A to seek out and consume DDT, perhaps by making DDT delicious or fragrant to A.
Unexpected consequence: organism B evolves to produce berries that are absolutely redolent with DDT.
This might happen centuries later, due to the DDT-phillic genes outlasting the presence of artificial DDT in the environment, or it could happen much sooner, or it might never happen. Hard to know. “Life finds a way.”