30 karma · joined May 9, 2021
This instantly makes a number of applications able to run directly on S3, sans any caching system.
Most roles that are run-of-the-mill “knows framework X” just enough to support naive clients or actually high/value teams, will indeed go to zero.
Which is amusing because 90% of the “AI” companies being started are run by people with no actual expertise, who probably are the most replaceable parts of deploying actual high/end systems.
Obviously increase supply (fracking, pipelines, offshore drilling, etc.)
And create more alternatives (actual cheaper alternatives, like nuclear power plants) as a substitution effect
Thanks to brilliant management, like handing out a billion dollars to some google exec who knew nothing about crypto and shipped a product no one uses, Coinbase is now indexed to the average price of the shitcoins it consistently lists.
https://python-zstandard.readthedocs.io/en/latest/
https://engineering.fb.com/2016/08/31/core-data/smaller-and-...
The compression ratio and decompression speed just blow everything else away.
Custom metrics Multiple phases/splits Remote evaluation Human evaluation Evaluation in Environments
The actual rankings are: #1 Kaggle #2 DrivenData
Honorable mention but poorly managed: AiCrowd
No one else has any level of funding to incent performance.
You could also accept that Faangs are full of slackers and coasters--stop tying your self-esteem to your work or your boss (who is actually your adversary, squeezing as much work from you for the lowest price). Do enough to be of interest to other teams there, have a few references and savings and just assume you will be let go, and focus on improving your life and your next opportunity.
"Machine learning projects – if ML is being attempted at all – are in early stages, using traditional methods that are best-suited for high-RAM CPU rather than GPU SKUs (ex: scikit-learn and clustering approaches)."
The idea that the machine learning being done there is in "early stages" is laughable, given the prize pools and sheer competitiveness usually move well past existing SOTA--usually moving forward benchmarks on Google's image classification and labeling (!) benchmarks and other areas where enormous teams can't match the top few.
Part of what you're seeing is the mass of survey participants, who show up to fork notebooks and fake ML skills, are the people who haven't exactly established themselves in the field (https://commons.m.wikimedia.org/wiki/File:Survivorship-bias....), while the top-end is too small a group to fully characterize with low-powered clustering.
Even five years ago, most real-world MLEs knew how to use AWS, were deploying actual machine learning models (granted, more primitive than today's methods), and I'm wouldn't be so glib about calling anything "early stage" even then given the raw business value it provides.
Is the world a better place because Gates used blatantly illegal monopolistic practices to stifle innovation in the OS space? Or because Facebook is willing to engage in dirtier data-sharing, growth, and addictive product design?
Whatever one thinks of those, Amazon created a market for the long tail of books, media, and then goods that simply didn't exist. And then followed it up with one of the most important platforms in the history of the tech and startup ecosystem.
He's done more good with those than she'll ever do by giving his earnings to whatever is most fashionable at the moment.
Python is slow, but PyTorch is fast, GBTs are fast, Cython is fast, Pandas and Numpy are fast (and even faster libraries or even basic joblib code can parallelize these).
Anything that needs to be fast either is or can be made fast--and most compute in data-intensive applications exists inside these optimized libraries anyway.
Good for her for marrying well, but let's not pretend this is about anything more than her desire for attention.
As much as I enjoy this article, low-end VCs are an enormous drag on a company, board seat or not. Constant bad ideas, short-term thinking, focused on how the business "appears" rather than performs, because they're in the optics and ego game rather than producing returns for their LPs.
Choosing very carefully who you let into the circle that is your company is one of the most important decisions you ever make.
Taking money for pure prestige of the firm, rather than an aligned and intelligent GP for the deal, or for highest valuation that has a 1-2% total dilution difference, is a bad idea given they can subtract way more than that by being a terrible investor.
(The best VCs like Founders Fund just admit they add no value, which saves you the worry of too much meddling from those who, very mistakenly, think they can.)
Pharma profits, big tech profits, and nurse tik-toks must take priority over disease prevention.
The quizzes weren't quite a fit for ML, nor was the interviewer or job density there for the domain, but the immediate half-hour call with the promise I'd passed a technical screen was extremely efficient, even if Triplebyte lacked any basic level of market liquidity.
TLDR: More or a whiny rant than anything resembling facts or data
100% of the 17 Randomized Controlled Trials (RCTs) for early treatment and prophylaxis report positive effects, with an estimated improvement of 73% and 83% respectively (RR 0.27 [0.18–0.41] and 0.17 [0.05–0.61]), and 93% of all 28 RCTs. These studies are tabulated in Table 1. The probability that an ineffective treatment generated results as positive for the 55 studies to date is estimated to be 1 in 23 trillion (p = 0.000000000000043)."
Just another reminder that the sole purpose of the FDA is to juice Big Pharma profits (when it's not busy juicing the profits of big food, to further help stimulate big pharma demand)
Footnote: the reason Big Tech also conspires to block discussion of this is the massive amount of spending they receive to market the blockbuster drugs that never have much efficacy and always seem to get recalled within a few years for actually being deadly.
Of course the virus most likely came from the lab, and had been engineered to be particularly effective against human cells. The burden of proof is one anyone else to show how it possibly evolved a furin cleavage site and many other odd sequences not seen in naturally occurring coronaviruses. And don't forget to explain why the outbreak started right next to the only lab doing gain of function research on coronaviruses, rather than near the bats thousands of miles away.
The dark truth is the "public health establishment" is happy there was an outbreak, anything to pretend they're the least bit important (they're not, and they got every single decision wrong during the pandemic).
But when ML works--Uber forecasting demand, Instacart optimizing shopping paths, delivery service route dispatching, and an infinite number of other real-world use cases with double-digit process improvements--it's extraordinary.
Sales forecasting, fraud prediction, modern search, and more are all powered by ML, and throwing a few stones at bad business plans, bad marketing, or over-marketing doesn't change the fact that those things are bad precisely because they're not that good at actual machine learning.