Well, you've picked a cheap product that has to work reliably in a very unforgiving environment. On top of that, it's a market that has a high saturation with plenty of alternatives. Also, it is really hard for a new brand to sell reliability and trustworthiness. New startups are better off selling innovation and unique solutions.
Personally, if I wanted to make and sell a product all by myself, I would look for a problem with no readily available solution, ideally one I have experience of, in an underrepresented market. Sometimes it's an unavailable interface between this and that, sometimes it's a new and better way of doing something, sometimes there is hardly anyone making a thing. Whatever, you should find a niche in the marketplace where a fresh product has a chance. Ideally, one where you can charge a reasonable price per unit, so it's worth your time. Incremental advances on existing products are a dead end for a new startup.
Back to the question of if it's possible, though. Yes, you can design circuits on paper. You can get free PCB design software and make your Gerbers. You can get small boards delivered for £10. You can solder them by hand. You can get a free 3D CAD program and make a Step or 3MF file, so you can 3D print enclosures at home. If you want something mass-produced and made by other people, then first get £100,000 in funding.
So while it's completely possible to be a 1-man band these days and get something out there that might get you some funds to do things properly. I would pick a different product, though. You said it yourself, "The manufacturer wants you to throw the light in the garbage and buy a new one" and for the average consumer, that is what they expect. They would rather spend 50% on a light they have to replace every year than 100% on one that will last 5 years. The manufacturers know it too.