As somebody who actually tried this — not much. You’ll need a lot of tacit information on how (thermo)nuclear weapons can be constructed, the feedback from actual engineers and experimental results, etc. I can design e.g. an implosion simulation with some open source CFD codes, but how can I verify my results without access to (really expensive) explosion testing rigs and measurement equipment? Thermonuclear weapons are a whole another story: the experimental equipment for testing radiation compression is prohibitively expensive. So, thankfully, a working nuclear program is still only possible for nation-states and extremely sophisticated organizations.