How to Begin as a Researcher (or Entrepreneur)

I was really touched by all the encouraging messages! It truly means a lot to hear from so many friends, students, and fellow entrepreneurs. If you will indulge me for one more message, to start 2024, I wanted to share a couple of thoughts gleaned from my tenure at Google and my previous startups. These are the same things I share with the students I mentor and the companies I advise, and I hope they prove helpful to someone on their journey.

• Embrace the "fail fast" mantra. Experiment, explore, and don't be afraid to stumble. But just as importantly, learn to recognize when things aren't clicking and move on. You might revisit that project later, or you might not. Personally, there were some problems in optimization that bothered me in grad school that I didn't come back to until 4 companies later and 26 years had passed. Most, however, I simply realized weren't as important as I initially thought.
• Perfection is way overrated. Don't let the quest for the ideal idea, opportunity, or product paralyze you. The most successful people I know took action, even with an incomplete plan, and discovered their “amazing thing” along the way.
• Early in your career, put down the books (perhaps a little too early for comfort). Read enough to grasp the basics, then get out there and build something, even if it's just a prototype. The hands-on experience will yield insights that no amount of reading can provide.
• Science and engineering are both incredible, but they're not the same. Building something is incredibly rewarding, but especially in this age of rapid innovation, understanding why something works can be the real challenge. In my own work, for AI and large language models, unraveling how knowledge is represented could lead to major breakthroughs in training times and data utilization. Perhaps even using AI models to analyze and summarize other models is the key to unlocking their full potential.
• 2 > 1. Work on two projects: the one you have to and the one that keeps you up when you really should be sleeping. You might be surprised how often the latter ends up being the most important. 

Finally, a personal philosophy: create, don't just consume. We're bombarded with content—shows, art, games, books. Think about how few of your own devices encourage creation versus just making consumption easier. Though the world pushes consumption, I urge you to create. Whether it's art, writing, or even just tearing up old magazines (like this, or this, or this), just create!  

If you've read this far, then I already know a bit about you—and I bet you have much to offer the world beyond your technical prowess. Moreover, I bet you underestimate your own talents. Remember, except for the very rare lucky few, for most of us, success comes from learning, practice, trying, retrying, retrying, and more retrying — perseverance.

Thank you again for your incredible support,

shumeet