garn810

Hey!

Yes I prepared, classical grind on leetcode, easy/medium (no hards), Neetcode 75, and etc

So, it's not like I ignored the preparation step, it's like that there's a specific "game" and "format" they're waiting for.. Because regardless of the fact that I'm choosing correct data structure or finally implementing solution (sometimes semi-optimal, but working), it doesn't count that I'm vocalizing the logic and explaining how I think. Looks like they're looking for perfect execution

But again it was just my experience, might not be representative

I don't mean to sound depressive, this whole step just frustrates me (I'm 10+ years in the industry, working on high load backend systems)

I failed every LeetCode style interview I've ever attempted. I mean, I'm lucky to even be filtered for the interview nowadays.. However, I failed not because I couldn't solve the problems. I solved them (honestly, without any AI). I just rarely solved them the way the interviewer wanted.

It's about time to reread books by James Hollis

"Under Saturn's shadow" for instance

Actually

Quickly they'll find a way to commodify the tech

Same happened with electric cars (modern Teslas are joke in comparison with BYD). Same will happen with AI

AWESOME!!!!

The most overlooked event of the year started in late August and is running as you read this: the Marin project out of Stanford is pretraining a 535-billion-parameter model with every metric streaming to the public in real time. It's the largest fully open, live-streamed frontier training run anyone has ever attempted.

Yet people sell vector DBs solutions as if it's a guarded magic knowledge

Whole LLM agent tool call with ripgrep gives 99% use cases right lol

American people are fed propaganda that Chinese are bad people with bad tech etc

In fact, it's often superior

Looks at cars, BYD, Geely etc...

Looks at neural networks benchmarks...

Ah, yearly I hate/love nix post..

I tell you what. Ever since I migrated my dev workflow to nix package manager (and home manager for system depa), my life became simpler like 10 times

I can configure any macos machine in like 20 mins (I use Determinate)

Yep. It's about time western world realized Chinese are not the "very bad guys under dictatorship"

The US is now doing a softer and broader version of the same thing to trust-based export sectors. It’s not the same method but! it is the same mechanism. The main difference is that the US damage is more reputational than structural, so it could be reversed faster (only if policy stops telling customers that dependence on America is a political risk)

Why's everyone selling themselves as "AI-native" these days?

Spacemacs is kind of bloated and easy to break with custom packages which are not part of original build

I've been using GLM since 4.5 version, only occasionally turning to Claude (because of price reasons)

With a good harness and instruction set, frankly I don't see the difference

People should stop thinking "Chinese = cheap", and maybe read less US propaganda

Haha, builder=ownership without boundaries. Management cope for "we fired half the team and now want the survivors to glue the product together"

Academia always been full of narcissists chasing status with flashy papers and halfbaked brilliant ideas (70%? maybe) LLMs just made the whole game trivial and now literally anyone can slap together something that sounds deep without ever doing the actual grind. LLMs just speeding up the process, just a matter of time how quickly this shit is exposing what the entire system has been all along

the marketing department of "Empire" runs deep, even on Hacker News..