I don't know that it's sexist per se, but if you use it to predict the outcome of future interactions (which I am assuming you do since you actually posted it here), then that is an example of bias. IANAL but probably actionable in the US.

So if your remark is, "In my limited experience I've worked with only 10 female engineers and the eight good ones all wore dark clothing and wear glasses. That's a funny and unlikely distribution!" then you're good.

If your remark is "I can determine if a female engineer is any good based on whether or not she is wearing dark clothing and glasses" then you probably have no business working in the field at all.