Short answer: No, the 3/5 compromise did not alter census data. We shouldn't expect to find enslaved people to the tune of the extra 2/5 if we hopped in a time machine and went back to count ourselves.

**Slightly Longer answer:**The first census took place in 1790, after the Constitution went into operation. So there's no "before" as such for the national census data. Some colonies conducted individual censuses before that, but usually when one asks about censuses one means the decennial federal census. Constitutional rules aren't an issue for statewide censuses conducted separately. The national census has counted slaves one for one since it began.

Doing it by counting each slave as 3/5, rather than just doing the math on the totals, would have required the census takers going around with their ledgers to not "3/5" instead of "1" and someone would have had to do the sums. That might not sound like a lot of work, but consider that it had to be done by hand on both ends. Far easier to keep ordinary running totals and do the math once per state.

On page 2 of this PDF, you can see slaves totaled up for 1800. It's all nice, whole numbers. While it's theoretically possible (I think. Not great at math.) that counting each individually as 3/5 could have yielded whole number sums, I checked the first entry (New Hampshire) to be sure. Eight is not evenly divisible by 3/5. There would have to be 13.33 slaves in order for the math to yield up eight. (Unless I messed up with my calculator, anyway. If I did, everyone is invited to point and gently laugh.) While you could count a slave as less than a full person, and the nation did for certain purposes, the individual census takers did not. The same issue exists with Rhode Island's 380 slaves.

This is consistent with returns I've used for later census, in particular the 1860 census. You can get the files of those here. (Be aware: large, zipped PDFs.) Down to the county level, they use whole numbers. The Maryland returns (PDF 1860a-08, page 5 in PDF pagination, page 214 in original), for example, have Table #2: Population by Color and Condition. The first three counties listed have slave totals as follows: Alleghany (666), Anne Arundel (7,332) and Baltimore (5,400). Results are also separated out by sex and there are separate entries for "Black" and "Mulatto". The "Free Colored" columns immediately adjacent and "White" just a bit further over on the left, are all also whole numbers.

This is consistent with the 1850 census, which I've also used occasionally, also down to the county level. Microdata (Individual pages or entries in the census books, laid down by name of head of household and in individual tracts, etc. Some of this has gone missing over the decades.) is something I don't have access to presently, but I have seen scans of entries there in the past that counted slaves down to household level in strictly whole numbers.