It seems mad that something like "ingredient parsing" could ever be a business; on the other hand, a decade ago I was sharing a house with a PhD student who was trying to do effectively the same thing with chemistry papers to extract relevant formulae. Worked well for clearly defined things like "1,1,1-trichloroethane", less well for colloquial things like "formaldehyde" or, god help you, "lead".
(Is a "lead researcher" the head of your team, a person researching the properties of the metal Pb, or a researcher made out of that heavy grey metal? This is practically an unsolved computational linguistics problem)
Lead (/leed/) researcher - most senior of your research team.
Plumbologist - a person researching lead (/led/).
Plumbous researcher - researcher made primarily from lead (/led/). Probably an AI that deeply resents the materials-based engineering flaws in its processor core and robotic reality interface.
Lead (/leed/) plumbologist - most senior of a research team investigating lead (/led/).
Plumbous plumbologist - aforementioned AI, but dedicated to self-improvement.
Lead (/leed/) plumbous plumbologist - most senior of a group of lead-constructed (/led/) AIs, investigating the mystery of why the human progenitors chose such a bad metal for building robots.
(Is a "lead researcher" the head of your team, a person researching the properties of the metal Pb, or a researcher made out of that heavy grey metal? This is practically an unsolved computational linguistics problem)