Even if the other parts are harder (not sure if thats true. The reason proving things work is hard is because most initially promising candidates dont actually), it doesn't follow that every single vaccine candidate works.
Early vaccines were made in the scientific equivalents of modern basements and worked adequately well. Influenza vaccines come out each year like clockwork. There are plenty of vaccines which just aren't rocket science, including for virtually all major historical viral diseases which wiped out or crippled large numbers of people.
And then there are ones like AIDS.
With COVID19, as far as I can tell, virtually all reasonable vaccine candidates DID work, and it was mostly a question of how quickly and how well. I'm much more excited about a 95% effective shot than a 60% effective one. The Moderna one was developed, quite literally, overnight. Once Moderna had the sequence, they had a (later proven working) vaccine candidate more-or-less instantly, in January, long before they'd seen the virus live, let alone had a way to check that it worked.
Did you see many /failed/ COVID19 vaccines? I saw a few /worse/ vaccine candidates. A couple failed in some populations (e.g. Sanofi worked well for younger people, but not older). Some didn't have competitive effectiveness (e.g. Merck, presumably, but they didn't release enough to know when I last looked). Some had side effects (UQ/CSL would cause people to fail AIDS tests later). Etc. I think all of these will compete for some of the logistics chains as the winners, so it doesn't make sense to push with /less effective/ vaccines.
My general impression is that COVID19, at least unmutated, is not hard to make effective vaccines for. There are questions about longevity and mutations, but for now, I can see the sticking-peptides-up-nose approach potentially working. I mean, compare that to early vaccine efforts like variolation (which worked), or the whole flu of 1918 vaccine effort (which didn't).
Disclaimer: Not an MD or biologist.
The hard part isn't making a vaccine, it's making one that's safe and reliably effective. Using 19th century techniques will kill you an unacceptably high proportion of the time (well, unacceptable unless it's something like widespread smallpox or, debateably, covid) and also often won't work.
I mean, it might be driving with the emergency blinkers on the whole time, at 10 miles per hour, and flashing the horn, or crazy stuff like that, but making a basic self-driving car just isn't out-of-scope for an amateur.
DARPA urban challenge was 2007, and 6 cars finished an urban course, with the fastest averaging 14MPH, and most developed mostly by students, with professor support. Today, I can get a lot of that stuff CoTS, open source, or read papers about what they did.
I couldn't sell millions of self-driving cars, because of edge cases, manufacturing, or what not, but it's not fundamentally different from biohacking. Technology moves forward.
Disclaimer: I'm not an MD, biologist, or similar, but I am pretty good at all the stuff that goes into a self-driving car. I'm not claiming a /typical serious amateur/ could build one. I'm claiming /I/ could build one.