Breaking In

Jeremi and Luca meet founders at tech events in NYC and SF

Welcome back to Jeremi and Luca’s Newsletter, an update every Sunday from two friends connected by a relentless desire to learn.

---

Luca: Interested People

A week ago, I joined one of my dad’s colleagues at Tech Pickleball NYC—a weekly event in Brooklyn with coffee, bagels, and pickleball, but more importantly, with interesting people.

It’s a cool and simple format: get smart, motivated people out on the courts together. And so far, it looks like it’s working—somehow, it’s free of excessive signaling.

By that, I mean to say that people aren’t showing up to say they went to Tech Pickleball, and they’re not showing up purely because there are beneficial people to meet. They’re showing up because of high quality peer-to-peer interaction.

My experience at Berkeley has been that there’s a lot of this sort of signaling. Clubs or events will be made with the right intentions, but they’ll eventually be overridden by people who “join to join” rather than to learn or contribute in a meaningful way.

Last Friday, I got coffee with a Berkeley student who was visiting New York for a few days. He told me about how his AI entrepreneurship club recruits via outbound rather than inbound. He’ll be on the phone with the smartest, most curious people he knows, trying to convince them to join the club. And it’s been working.

I also talked to Jasper (a Berkeley friend) this week, whose role at a VC firm has driven him to organize dinners and poker nights for student founders to meet and share ideas. He too is building cohorts rather than waiting for them to come to him.

***

As I was writing this newsletter, I looked at a list of running notes from the past few weeks, where I found myself writing down, multiple times, various iterations of “look into new clubs at Berkeley” and “find more people at Cal.”

So, I guess a goal for this fall is to do more of that, to figure out how to find really interesting people or bring them together. I think it’d 10x my experience at Cal.

Anyway, speaking of spending time with smart people, we ended the week on our apartment’s rooftop with my roommate and two of our startup colleagues who are also still in school. We’re having a good time this summer.

Subway home and rooftop with colleagues

---

Jeremi: Becoming Technical

I think I’m close to being useful.

On Wednesday, I attended a pre-ICML mixer (thank you Jed for inviting me!). ICML is a conference for Machine Learning. I spoke with a founder of a Robotics and AI startup, and I told him about the project I’m working on. He grilled me, in what sort of felt like a dance—he would critique, and I would respond with a counterpoint or a follow-up question. I held my own for about 90% of the time.

Later, I asked him if he was hiring. He said (and this is not the first time I’ve heard this) “We’re only hiring for top AI talent. But we might have space for a smart intern next summer.”

Recently, in the news: Meta poaches three OpenAI researchers, likely for hundreds of millions of dollars. There is a small cohort of the “top AI talent,” and they’re extremely sought after in the valley. But it’s hard to break in.

The next day, I went to a “Deep Tech” mixer and spoke with a founder of another Robotics and AI startup. I asked him some questions about their product. He referred me to their head of AI, and told me that they were hiring.

The networking acts as a good gauge of where I’m at. I talk to technical founders or researchers, and I see how confident I feel.

In both of the previous cases, I can tell that I’m almost useful. I understand some of the problems that these companies are dealing with. I understand the different solutions that are investigated, and what their pros and cons are.

But if I was thrown into the middle of one of these startups, I still think I would struggle to swim. Part of it is filling in the gaps of knowledge that still remain, as there are a lot of techniques I’m ignorant of. But a large part is the concrete implementation: taking these abstract ideas, and realizing them in real-world scenarios where customers have a lot at stake.

That is my goal this summer: go from almost useful to truly useful.

I appreciate the reading group my housemates and I have started—it will help me fill out my gaps of knowledge. We’ve read the paper on diffusion models, and have plenty topics to go into (Residual Networks, Vision Transformers, Flash Attention, and more).

The other side of the coin is working on projects: get to the concrete. Get my hands dirty, jump into the implementation, and get some results. In these last two weeks, I’ve sometimes felt like I’m making progress on this goal, and other times felt like I’m running in place.

I’m probably further away than I think I am. The top of the hill always looks closer from below. Still, I’m excited. I can feel the progress, and it just makes me want to work harder.

Walking through downtown SF, hacking on laboratory robots, lunch with Jacob