runeblaze

To be honest I am pretty sure 95% of the people like play games and ride bike more than just coding.

Yep my phrasing was not the best :X. You might know better ("couple years"). However my anecdotal observation is that the end result is that fintech drifts FP (heavy concurrency + immutability, ledgers and all) than your "average" tech. I genuinely don't know why, maybe correctness guarantees and ledgers are just so important.

While I agree with the boredom premise (I was also bored!), the practicality of Clojure was also very important. In fact if I were to deploy something to prod today that is FP, then I either use Scala or Clojure.

Fintech benefits from concurrency, immutability and all, so obviously there is much more nuance to "why fintech uses Clojure".

I write from a new grad perspective, but as said, put your well-being and the well-being of those you care about above all.

Meanwhile don't beat yourself up if you are young (bonus if you just relocated for work) and spent too much time at work or feel "loyal" to your employer. Wind down, of course, but don't beat yourself up.

But jsonnet is Turing complete…?

To be honest my interviewers couldn’t sound less interested when I told them about my thoughts on Camus' Caligula and my love-hate relationship with Livy's History of Rome when I applied for jobs, same when I applied for PhDs.

--------------

Similar things happened when I try to quote Dijkstra and "Out of the Tarpit" during coding interviews. I then started to quote Uncle Bob and they start to understand more. I am not sure people care about reading. Mind you this was the new grad job market.

> Is Lanczos an example, or the current best option

What’s the best (computationally not that more expensive than Lanczos) option?

Edit: also some CV researchers write like that (the Netflix writing) — bicubic is like a flag in opencv that they just use. Probably those researchers were much more preoccupied with the researchy problem than actual wide deployment, which is what many researchers do

They are good at (combining well-known, codeforces-style) algorithms; often times I don’t care about the syntax, but I need the algorithm. LLMs can write pseudocode for all I care but they tend to get syntax correct quite often

Personally reading this thread I just realized that devs encompass so many areas. Requirement for senior engineers working on practical formal methods, real-time systems, Linux-kernel hacking, or those writing CUDA kernels all day are all gonna be different. System design is only for a broad kind, but not for all kinds, of senior engineers.

In the US, note that for many foreign students, PhD is a (in many cases, much faster) pathway to permanent residency compared to your standard PERM/H1B things

I want to vaguely highlight that they are not that competitive. Headcount is not that great; there is usually a citation/research impact floor; the ivory tower "oh our advisors are friends" thing likely also applies.

Research Scientist positions are also embedded in product teams in Google IIRC, certainly less "prestige" than DeepMind though, and probably easier to get into

(Also for the broader audience: Harvard here probably carries less weight than the likes of UT Austin)

Useful information for HN from the original Google blog-post:

> First, the 2018 analysis flagged one particularly large job code (Level 4 Software Engineer) for adjustments. Within this job code, men were flagged for adjustments because they received less discretionary funds than women.

In other words, "some jobs" here is mid-level software engineers.

I am literally guessing at this point — but can they mmap those files?

(Or I mean can you shard the files/store the files more efficiently)

I think we all agree this feature seems broadly of use, and given that presumably a professional full-time 1P team was behind this feature I am gonna use this product over the other implementations

I mean you can treat it as just a general pseudocode-ish implementation of an O(n) find_max algorithm. Tons of people use Python to illustrate algorithms.

(Not to hide your point though -- people please review your LLM-generated code!)

That’s your H100 and A100’s GPU memory, not too sure but I imagine many AI practitioners would not find that a weird number

While I am sure the foot soldier quality is important, we ought to put the burden on leadership a bit more. I am not sure AMD had a better talent pool (I don't work in the industry so I don't know!) ten years ago. Culture is predominantly shaped by those already in power -- it sounds like they need a change in culture.

I am sure Intel has enough cash to hire some good talent (like just offer talented people next-level salary at $FAANG), the problem is deeper in the hiring pipeline -- convincing people of the budget and actually scouting and retaining good people

Maths also mean different things. Your average number theorist or algebraic geometer will most likely not touch statistical techniques day-to-day. Reading this Anthropic article was helpful because I am constantly catching up on my statistical background

My limited corporate experience is that you fluff feedback to the less effective leaders so that you can still influence them. The more effective leaders you talk to them directly and team up with them to accomplish great things together. It is a matter of different strategies.

Sometimes I write Rust with Arc and Rc sprinkled everywhere because I just need a dialect of OCaml.

Smaller companies need to outbid Google for talent, especially if they carry less stability. I imagine there are quite many who went like: leaves Google for great cash package at smaller company/start up -> smaller company goes belly up -> go back to FAANG

For machine learning (un)fortunately, lots of the stack runs on Python. Lots of ceremony was done to circumvent GIL (e.g. PyTorch data loader). “Reasonable performance Python” I imagine is actually something in huge demand for lots of ML shops

Just to add -- sometimes I use Redis when I don't have a trusted impl of HyperLogLog or sorted set (and I have a vague suspicion that I am going to do IPC later -- so not worth it to wrap my own HLL).

One thing I disturbingly realized, after I tried corporate life, is how much misogyny and racism hides below polite discussion, and how invisible these things can be even to direct leaders and customers

Thanks for the feedback if you are talking about me ;). I frequently get suggested to put more punctuation...

Trust me when I play minecraft, the moment I realize I don't have inventory sorting installed I quit the game.

Same here. Also Factorio AFAIK has no good copilot. Sadly the best LLMs are not fine-tuned on Factorio so my productivity takes a massive hit when moving from side projects to Factorio.

Half jokingly, MrBeast is a decent source of learning other languages given (1) how children friendly the contents are, and (2) how many dubs/subs the videos have. I particularly like the Takeuchi Junko Japanese dub.

I think we are converging at an opinion. Internal actors can be hard to detect, and honestly there is a reason at places like Google interns are treated with heightened security checks (my guess -- they learned to do so after some years).

Btw one of the rumors has that it is even difficult to hire engineers to do training/optimization infra at one of those ML shops -- all they want to hire are pure researcher types. We can imagine how hard it will be to ask for resources to tighten up security (without one of these incidents).