As @WYSIWYG mentioned, I think HLA type is probably the best way to claim specificity. Then the glycoprotein (fusion protein probably) on the virus fits only in an exact MHC class II molecule. If you want to hit some valid heavy science jargon, I recommend saying the virus was made with recombineering, using a bacterial artificial chromosome (BAC), with a florescent marker to make the scientist's life easier (mush harder to work with viruses without nice markers).
Here is a methods paper that describes that exactly, and has the advantage of of a free published protocol as well.
My problem with this method (using an HLA type to target) is that if you're targeting MHC Class II, it's going to be hard to kill the victim quickly. Kill off all the cells would give a person AIDS, but not necessarily lead to a quick death. Targeting MHC Class I could get around this, but then you're not targeting specific cells (could be nerves, could be skin cells), and the person would probably mount a pretty good immune response. You'd probably have to hit the person with a pretty high dose to be sure you were infecting the cells you wanted to kill.
If you're willing to allow a two-step process, I think the more targeted approach would be to create an inducible latent virus. Take one that commonly infects humans like CMV, and modify it so that it was only induced out of latent infection by a specific trigger (tagged food for example). Then further modify it so that it generates a highly toxic protein when it comes out of latency, like botulinum, that will both kill the person very quickly and have the possible appearance of something natural. If you use food is a trigger, then health workers would eventually figure out were all these people were eating and accuse them of food poising with botulism. Whether that can help your character in the plot, I don't know.
Back to the creation of the virus, I would still use the method above, though it's important to note that botulinum is a select agent, and that proper BSL would be a serious concern. Have a hard time imagining it going down in someone's garage without them killing themselves. But if the person known how/has materials for gene synthesis (many, many labs do), then acquisition is just a question of making the gene for the published botulinum sequence.
It does occur to me that I might not should even post this based on it's ability to work...feel free to vote/comment to close if that is conciseness.