Works on problems where the method is not settled yet.
Why this one is hard to judge
Publication record is a signal but not the signal, because the applied question is whether it survives contact with a product.
What to ask for
Ask for a specific result they published or produced, and what surprised them about it.
Ask how they decided a research direction was not worth pursuing further.
Ask for an example where their result did not replicate, and what they learned.
Ask how they would translate a research finding into something a product team could use.
The mistake most hirers make
Hirers are drawn to research candidates with an impressive publication list but never ask whether any of it changed a real decision. Research that never meets a product or a deadline can become an expensive hobby. The useful researcher can explain their work to someone who will never read the paper.
What good looks like after 90 days
One research question answered clearly enough for others to act on it. A written account of what did not work, not only what did. At least one conversation where a research finding changed a product or engineering decision.
How we assess it
Against a rubric that is published in full, on evidence the practitioner supplies and a reviewer checks. Where something has not been verified, the profile says so.