Engineering is where things are proven, by building things using or that are proving scientific findings. For example, Randomized Controlled Trials are one of the final engineering steps for drugs, with partial engineering step being chemistry and process chemistry. Similar steps should be done but are often omitted for social sciences.

The exception to this is pure mathematics.

Example from biology: you found some fun cellular process model description. Fine, now you have to mess with it using engineering to prove it actually works in the way you described, is not missing crucial steps and so on. This for example happened a lot in gerontology - studies of aging. People latching on to small pieces of mechanism which actually are not very relevant or actionable.

Examples from physics are machines that use physics theories, such as quantum communication devices, particle colliders, clocks or nanostructured devices.

Examples from chemistry are evaluations of synthesis processes step by step by introducing various process changes.

Analysis is rather on the side, as without engineering first you cannot really proceed with it.

This may be debatable, but seems to accurately describe what most academics call "science". The more common understanding is what's described in the parent.

Often times when people say "trust the science" they're referring to results in the first category, but people take it to mean not "these results are preliminary and unproven" (and filled with asterisks), but instead "to deny this is denying empirical fact."