You cannot do everything

- 3 mins


I had an AI Summer. In June I went to my first hackathon, won, and when the judges asked us “So are you going to make a company out of this?” We started to seriously think about it.

Given that for the past 2 years all I had focused on was building AI at Uber, I was completely curious about what building AI outside of big tech looked like, so I started going to more AI Hackathons in San Francisco to learn more. Cerebral Valley made is super easy to attend hackathons, since they maintained a google sheet with every upcoming tech event and hackathon happening in the bay area. All participants had to do was navigate to their site, see which hackathons they wanted to attend, and apply to hack at them!

I attended around 10 hackathons and got a finalist award in around 5 of them. Perhaps one of the most fun programs I participated in during the summer was the buildspace nights & weekends program. It’s a 6 week course where you have to build your idea out from idea to the first 100 users. At the end of it, you meet in person in San Francisco for 3 days with all the graduates of the program to gain more users for your idea.

It was there where I was sitting, seeing many others excited about their idea, and being confused on what I wanted to do. I went into these programs with the purpose of exploring, but being in the program and seeing everyone take it so seriously made me feel that I should also - but that I wasn’t quite ready to take it seriously.

I was reflecting on the 4 or 5 different ongoing projects I had going. I was heading a very challenging data engineering inititive for the risk Machine Learning team at Uber. I was also developing machine learning models Uber needed for fraud mitigation. I also trained my own LLM that Adam D’Angelo (Ex-CTO of Facebook, CEO of quora, Board of OpenAI) voted 2nd best custom LLM. I was building a podcast generator with some friends, and additionally attempting to create LLama2D with some friends - a fine tuned version of LLama2 that was intended to make web agents more effective at surfing the web.

Every weekend day was filled with a hackathon or a meeting to catchup with work leftover from a previous project. Every weekday was spent working with my intern at Uber on innovating our data infrastructure to give machine learning engineers across Uber access to amazing tools for training data creation.

It was too much. I didn’t know what to do. So at the IRL event for buildspace, I sat there. I watched others. Familiar faces popped up and encouraged me but I was still totally confused.

Then I got the best piece of advice I’d gotten that entire summer. Talking with Shai, the cofounder of everything bagel, he reassured me that this was completely normal for 20-something year olds in technology. He told me that he ran across the same problem and one simple solution helped him figure it all out. It was three words. He said the solution was he simply “Took a break”.

That same day, I logged into workday, requested 3 weeks off (At Uber we have unlimited PTO so you just pay for time off with guiltyness), and marked my calendar as out of office for those weeks. I grinded to finish work for a conference we at Uber were presenting at in September, and dipped out after my intern had left and the conference was over.

I visited family and friends across the country for the next 4 weeks (the final week working for Uber from the New York City office)

What Shai told me spoke to me as well. “When you take a break, everything that’s not important to you will drift into the background and you’ll be left with only the important things on the top of your mind when you get back. Do those important pieces.”

I’m now back. I’ve now filtered out what’s important and not important to me. I’m excited to share what I’ll be up to going forward.

Chris Settles

Chris Settles

Technologist, ML for Fraud @ Uber.