Academic Conferences
"I'll be on Whova before the conference starts and cold-message random people like a cringy salesperson about why they have sushi in their interests. And that usually turns into a great conversation"
I am continuing my tradition of going to ICML without presenting anything!
In preparation for the conference, I have been packing my bags and filling my schedule with luma events. I also want to mentally prepare to get the most out of it and ask myself: why am I going there?
"I'll be on Whova before the conference starts and cold-message random people like a cringy salesperson about why they have sushi in their interests. And that usually turns into a great conversation."
Goals
So far, I had different goals for each conference.
ICML24: I wanted to get an overview over the field, put faces to the papers I've been reading and learn how other researchers operate
AAAI25: I presented my work and tried to be a bit more active, e.g., I 'organized' a somewhat random evening centered around privacy people, my poster neighbours and people I met from a german-speaking meetup. I put 'organized' in quotes because all I did was lead them to a bar that didn't have any space and then we luckily found another place nearby.
So what do I want from this ICML? I think my main goals are:
- Reinforce my already existing relationships, e.g., with the academic family
- Get a sense for what pathways after the PhD could look like (you can probably find me at the co&co x Vector Institute AI career compass!)
- Find relevant and inspiring work
- Refill on inspiration and excitement!
I think you can do so much at a conference: find collaborators, find your next job, sell a product, advertise your work, get involved in the academic community, get funding from VCs, meet your research idols. Narrowing it down helps me focus.
Turning goals into actions, or, why I am skipping presentations
Goals without actions are a weird kind of misery, so I'll have to turn my goals into actionable things.
First, I skip most if not all presentations. Obviously there are great presentations and people put a lot of effort into them. I don't want to deprive them of their audience, it's not fun to present in an empty room. That said, they are often not worth the time to me. They are the least interactive/social part of a conference and you can just watch them online later! Unless you really want a selfie with the speaker, like this crowd around Andrew Ng (bonus challenge: try to find him lol). In that case, you need to attend I guess ...

You should decide on a case-by-case basis, but given the option to either attend all presentations or skip all of them (and instead, say, meet with people), the latter has the higher expected value for me.
Building relationships
Getting to know people on a personal level is, to me, the greatest perk of going to conferences. And I don't mean walking up to people and talking strictly about work. They write papers and have blogs of their own, you can read that stuff up. I mean getting to know their personality and what drives them. Like getting to know what a down-to-earth nerd (in the best way![1]) Nicolas Carlini is (okay maybe that one's obvious, since he writes tic-tac-toe in printf statements for fun) or how much our incoming faculty Andrew Lowy loves dogs and wrestling
So usually I'll be on Whova before the conference starts and cold-message random people like a cringy salesperson about why they have sushi in their interests. And that usually turns into a great conversation, with some of those turning into lasting connections.
Also poster sessions are a must; they are all about talking to people. I don't really like the mini presentation people usually give you when you come up to their poster, especially if there is a crowd. But if there are few people and you can really interactively talk to the authors, that's the best-case scenario for me.
I have been told by Nicolas that maybe I should change this to 'in the worst way', but I leave that decision up to the reader. ↩︎
Ask people about them.
Maybe you're done answering someone's questions at your poster or are just talking to someone, it's easy to forget that you can then also ask about the other person! Curiosity drives us researchers, so wondering about the person you are talking to will come natural to many. If it doesn't, you don't have to believe me. There is countless advice, like Carnigie's well-aged how to win friends and influence people or Alex Hormozi's sales playbook. They will more convincingly tell you that showing genuine interest in other people will generally help you in networking (to whatever goal). I don't think we should go about our days like sociopaths, methodologically decomposing and min-maxing our every social interaction, but keeping some of the benefits in mind can motivate us to go outside the comfort zone!
Selective Knowledge
Conferences are packed with content about really important stuff and filled with smart people that could legitimately 1000× that content on the spot if you talked to all of them. So I got very picky about what I care about and who I decide to get more knowledge from. This doesn't mean to strictly ignore everything outside your field. I don't do biology, but a talk on AlphaFold was very inspiring to me.
You can draw inspiration from work in completely different fields. Either their ideas are just directly applicable to a problem you have as well, or you completely misunderstand something and creatively find a solution for yourself. Either way, sometimes surprising things can emerge from seemingly irrelevant info.
I don't really have a cool technique for selection here, except trusting my intuition on what feels interesting. After a while at conferences, some things feel unusual. Be it through a really great presentation or authors that spark a shared excitement in me. And I like finding those as my anchor points in the vast sea of options.
TL;DR
You can overthink everything. But with some mindfulness of my goals on the one hand and the possibilities on the other, I think navigating conferences is mostly intuitive. See you around!