Senior AI Research Scientist here. I feel really constrained in my actions while working in big tech. Can’t do much about it. Need approval here, need approval there, need to consult legal, need to buy license. Everything takes weeks and months (sometimes even a slack response). Nobody really cares about doing great things. Everyone just consistently do their part of the job for years and don’t want to jump above their head. I understand that most of the people are kinda old for that shit, they have their children, wives, pets, mortgage, visas and so on. I don’t have any of that. I’m in early 20s, haven’t even finished MSc yet. So recently I started thinking about doing something by myself. I don’t have much corporate “team lead” experience (have it on my pet projects though). So objectively speaking, I don’t have any proven management skills at the moment. So maybe I should work more to get more experience. From the other side, a lot of people making up these excuses for their whole lives, like getting MBA and other useless scam. Some people say that you need to be aggressive and decisive to get success. My idea is to make AI research startup something similar to OpenAI or Anthropic, but work on a bit less general tasks, specialize on my country’s language. Get money from government and funds. Invite people that I know and which I believe have bright minds (though working at some shit companies for low salary). What do you think?
How are a senior AI research scientist without finishing your masters. We don’t have a single research scientist without a PhD.
Go for it. By the time you get management experience, you’ll have a partner, kids, pets, mortgage
If you can find funding go for it. If it succeeds you are set, if not it's going to be a fun experience. LLMs are an expensive sport, how do you consider positioning your startup against openai & friends?
Well it’s obviously hard to compete with OpenAI, because they have enormous investments, very smart people and Microsoft data centers. Therefore, the work should be related to other field. My idea is that OpenAI and friends provide great products for English, but it’s possible to create a more specialized models for less common languages. The funding is also a big question. Previously FTX was generous about giving money, but now it’s gone. Like I said, currently I’m reading about government funding for young startup, but I’m not sure about that.
When? Work on it on weekends. Do minimal work at the company.
Based
Now. There is no better time to raise funds for AI startups than now. It doesn't matter how prepared or not you are, the investment climate will be different in a few years, the land grab will be done. If you want to start an AI startup, do it now. VCs are handing out millions to people with just an idea. Personally, I think "OpenAI, but for some tiny country" is a shitty company since OpenAI will eventually scrape the same data, but if you don't have a better idea, you gotta fight with the army you have, and you can always pivot later, you might not be able to raise with basically no plan later. I'm not at Google any more.
**Get money from government** you have my attention