"He did a stupid thing, therefore he's stupid" is not ad hominem. Inflammatory, subject to a huge debate about definitions, perhaps even wrong? Sure. But if you expand it out reasonably (a sample intervening premise: "People who do stupid things are stupid."), the logic is valid, in the technical sense. To attack the conclusion requires attacking something else about the premises, the most obvious fruitful avenue being the definition of "stupid" in question.
(Although, spoiler, that pretty much ends in "I have my definition and you have yours, and by mine, yes, he's stupid, but by yours, no, he's not.", and as long as we don't mistake the map for the territory and argue as if the definition of "stupid" that is used by someone has any power over the real world, is a perfectly acceptable result.)
(Incidentally, I consider my sample intervening premise up there as stronger than I personally would use the word stupid. I like to think of myself as generally not stupid, but I've done stupid things. Still, with that premise, the argument is logically valid.)
Which, given the evidence, they unarguably are: Virtually every expert in the field, as well as other biohackers, think that what he’s doing is a bad idea, and based on dangerous superficial knowledge. The comparison to Barry Marshall (the stomach ulcher guy) is particularly unhelpful since Marshall was an expert in the field, and acting on a solid theory and with a contingency plan. But even his actions are still rightly called reckless.
All that said, I’m still impressed with Roberts’ self-experiment, and curious about the results of the long-term follow-up. But it is stupid.