To fool forensics, you need to defeat their databases - and databases are not magic, they have limited stuff in them depending on what's useful. If you just want to fool a forensic lab that focuses on criminal activity, it will have a database of human-made substances, and local biological substances, and lots of capability to pick up the stuff that points to human variations and actions - but they won't have every exotic animal ever, and may be stumped by some substance or specific composition that's common to a locale halfway around the world. If you pick something suitably exotic, or start messing with gene splicing from those exotic roots, you might end up with a puzzle that might stump them for quite some time.
The problem with that is, humans escalate - having stumped your local police-oriented forensics, they will escalate - to national databases, to scientific ones, to hunting down common threads (kinda-sorta-maybe like that is pretty weak, but if it's all you have...) and finding specialists. And for something as attention grabbing as aliens, this will be getting the kind of attention to keep escalating to those higher databases and more specialized scientists. Every database that fails to find a match will increase interest, which means more scrutiny for any small mistake you may have made.
Practically - you should take an interest in cutting edge biology - the stuff interested in studying bacterial cultures of thermal vents or sulfur pools or other deeply unexplored areas - ones where new discoveries, different species, and unstudied adaptions are likely to be found. You need contacts in the field - so you know what's going on, what ideas are being tossed around, and also physical access to sample stuff. Obviously it will be best if you and any co-conspirators can study and get samples without anyone knowing anyone has done so, but a good compromise might be being knowledgeable enough to be able to eventually throw investigators off the trail when they reach your level of specialization.
Then, play around with whatever you've got - maybe gene splice things that kinda work together and kinda don't... or maybe skip that, because artificial is probably easier to discover than simply exotic. After all, the techniques that splice genes together that we know enough to use, we also know enough to detect - but genuinely exotic fragments can be mixed and assumed 'damaged', or interesting substances simply harvested - something greasy for the base of your 'fingerprint oil', something thickly liquid for a 'blood' analogue, some strand of proteins for a 'hair' - such substances that can be muddled with traces of your gene-spliced bits so that they seem to work together.
There's several layers of materials that can be layered together to make a really surprising picture - DNA fragments, proteins, lipids, chemical reactions, trace elements. You don't have to make them work together seamlessly, you just need to have enough puzzle pieces working together to suggest enough of a picture and make your researchers fill in the blanks... and enough blanks to make the bits that really don't fit, look like the problem was the gaps.
For example, using a medley of sulfur bacteria DNA with some of the kind of oils found in (or useful to) a sulfur-rich environment, and traces of sulfuric acid at the scene, for example, can let traces of proteins or other structures not usually found in sulfur-environments (maybe from something seagoing, or from insects, or even vegetation) seem exotic and/or alien instead of cobbled together. Or if you want your 'fingertips' to have hexagon-patterns, use bits of DNA from species with hexagon-coloration, or a tendency to use hex patterns for building with (like bees' hexagon-shaped cells in honeycomb), or something.
Then - build your 'crime scene'. Carefully, and preferably with no real crime involved (the better to invoke scientific curiosity, rather than outrage, and to make it as low-priority as you can). Your robot hands gets a tiny smear of 'fingerprint oil', and touches or grabs what you need it to, maybe a few bits of other evidence like alien blood, skin cells, hairs (depending on your harvested material-alike's). It will be much better if you build a clear picture of what was happening (it stood here, grabbed that, knocked something over) as that kind of material picture is the strength of police-type forensics anyway.
Then destroy swathes of this evidence... wipe things down with sulfuric acid, 'attempt' to set a fire, basically assume your alien didn't wanna be caught - but that it 'missed things'. This will help explain why you've got partial prints, fragments of DNA, plenty of evidence of your oils, prints, and various fluids - but not a lot actually in a state to be tested, and plenty of reason for the tests to find gaps where things are missing. After all, the best way to hide that the puzzle didn't start with all its bits, is to make sure they're expecting not to find all the bits.
The more you paint a bigger picture, the better everything will work because people like narratives, they want the story to make some kind of sense - so you wait for external factors (meteor shower, or storms, whatever fits) and maybe use a series of 'scenes' to hint at a larger picture or provide frame for that narrative. One scene might be an anomaly, several as the 'alien' seemed to move in search of something, was maybe avoiding something, until whatever happened. Again, you don't need answers so much as hints that there might be some reason only partial evidence remains. The alien needed something, it crashed, it was looking for someone, it left, it died (and no remains because...).
They say earthworms and humans share about 98% of all DNA-structures, from having evolved in similar environments and having similar distant-ancestral goop. That means, something only a little bit different - say, sharing even 95% of our structures, is still pretty likely to get touted as alien. Sulfur pool or thermal vent bacteria might qualify - they're living in a very different environment, after all. Bonus points if you can incorporate things that seem mistaken or wrong until someone gets far enough up the scientific ladder to say, actually it works - it leads them in the direction of science-we-dont-know-yet instead of hoax, since a hoax is a lot more likely to use things people know or assume, than those scientifically correct but seemingly unknown. (extra bonus points if this is a discovery new and/or obscure enough that even scientists in the field might need to double-check if it really works.)
And even after that similarity has been tracked down, especially if you're using the really cutting edge of discovered species, you might be able to persuade them the similarity is about as coincidental to your aliens as earthworms' are to ours, instead of being muddled from them - that is, about similar environment and convergent evolution/adaptation, starting from an environment that is rare on earth, but must be more common wherever these aliens are from - your sulfur-abundant world, or one run on heat instead of light (from a thermal-vent themed selection).
If you're very careful, and have lots of resources, and a fair amount of luck - you might manage something, that fools them for a time (and if everything runs in your favor and that time is decades or a century or so, I would count that a success). In time, of course, technology will advance far enough to simply overcome your obfuscations, since you are limited by whatever tech you had to start with.
All due credit for inspiration goes to the Cottingly Fairies hoax, whose photographs could not be proved to be altered because what they manipulated was the scene being photographed, instead. Keep people focused on the bits that aren't a hoax (use real materials, for example, your bacterial DNA is biological) and the bits that are (location, robotic hand with hexagon-fingerprints) can get glossed right over.
