@sergey_science That’s the myth again, and in this case, I was talking about Berkeley students. Everyone knows a founder, first or second degree at most. Which makes it more achievable and dispels the apprehension one would otherwise subject themselves to.
YC & the SF VC World is often mythicized, people overthink it and get in their heads about it. The reality is much simpler.
When you see your friends, normal people with their highlights & lowlights, get into YC or get funded - the myth slowly fades (this is a good thing!) - You can afford to get our of your head! Because the evidence [your friends] proves otherwise.
Berkeley does a phenomenal job at nurturing talent, but we can't discount the fact as there's plenty of alumni & batchmates who've gotten into YC, it's just... normal. It's not crazy rare or whatever, everyone there has a friend that's in YC or done YC before.
They're no longer psych'd out because they see reality for what it is :)
Halfway through W27 interviews. Intense weeks for us and for the founders applying, and a privilege to hear how they plan to change the world. Many are still in school or just graduated.
There is one school that consistently stands out this cycle: UC Berkeley.
Great hiring is dependent on identifying alpha years before it's obvious. Qt'd is bad advice.
Their resumes are empty because they're young, not because they're not talented. Everyone in the valley knows that young elite talent (<21) isn't logo-stuffing their resume.
If Peter Thiel won that case competition, he'd be an assistant to some judge in the middle of nowhere. Don't condemn yourself to a life of working at bigcorp, especially when you're young.
If you're a diamond in the rough, keep building stuff, put yourself out there, and opportunities will find you. Trying to force a logo on your resume is antithetical to this, and will make you chase vanity that will diminish the total value of your career.
Unpopular take but if you’re really early in your career you should optimize for trying to get 1 impressive logo on your resume.
Yes, that might be an early-stage startup you found / join - but the safer bet is to bite the bullet and go somewhere bigger / already elite. Even for a few months.
Logos are often the fastest way to determine potential of exceptional ability without talking to someone. Anyone who tells you otherwise is lying.
I [get an agent to] batch merge PRs once I’m done testing and reviewing them. It’s much better at fixing conflicts as it has broader context, and prod moves discretely in all our repos (every few hours, instead of an autonomous software factory).
Unfortunately I’ve gotten used to CI being run a billion times (not worth fixing), but atleast it’s not on gh’s insanely slow runners. Depot has a great YC Deal and fast runners. Would recommend.
@kaushikimmadi Yep, people complaining is better than absolute silence! Most startups die because nobody uses them, not because their users are annoyed over an issue :)
@Aniopsace Yep, completely true. The world is such a big place that even in the most saturated markets, you can squeeze out some demand. There's basically no excuse to be pre-revenue for SaaS, unless there's a regulatory or capex barrier.
@Aniopsace Internalizing PG is great, but I'm not so sure about pre-revenue. Anyone decently competent will validate their idea as fast as possible, not doing so without a good reason is an anti signal
@VoxelPrismatic@teddywang86 full duplex voice models have occasionally unintentionally mimicked the user’s voice (a hallucination), audio tokens are definitely capable of carrying accents anyway. This looks like a full duplex and can likely do accents 🤔
Everything a Co. writes is written with a specific demographic in mind.
It’s clear who this writing is meant for, and it’s fair that it sounds illogical to everyone else.
David Sachs mentioned how the Claude constitution has weird stuff. Skimmed it and uh...
Guys when humans die they decompose. What cult bullshit is this obsession with preserving old model weights?
Welcome to the real world. Today we’re introducing FrontierEE Bench, the first benchmark to test frontier models on electrical engineering tasks at a physical lab bench.
Real-world engineering benchmarks should involve real hardware. At our lab, models control instrumentation, run experiments on physical hardware, and decide what to test next.
The best models scored 36.7% across a range of EE domains including mixed-signal circuits, power systems, & electromechanics.