If you want to do fundimental new discoveries like this, then a lot - there's not a lot of open source code out there - most of what's on github is just students re-hashing methods from > 5 years ago. If you want to push the boundaries the best way is to read papers, practice implementing yourself, and befriend quantum physicists and computational chemists around the world that you can ask when you get stuck.

Chemistry is an enormous field too, you would have to concentrate on a much smaller section, I have been self studying for about 8 years now, (+ I have 20 years software development experience) and I think I'm getting to the point where I have complete understanding of the current SOTA and am pushing at the boundaries -- most of the ideas I am implementing are coming out in papers weeks or months later, so I feel like I'm at the edge.

Self study requires an enormous amount of discipline though, which is partially why universities exist - because most people do not have the discipline to stick at something for 3-8 years, and I am doing it 7 days a week, almost every day of the year - because I just love this stuff, it's not tiring, it's exhilarating!