A lot of very smart people work in strange ways / with a lot of quirks (e.g. contemplating for hours and appearing to do nothing, while then suddenly having 100x output burst).
This usually makes them not a great fit for traditional corporate world, where you often have to fake
using Amazon in 2024:
- type in search term
- 90% chinese no-name dropshipping products
- go to alibaba / aliexpress
- see exact same products for 1/3 the price, even using same product images
how is this a 2 trillion dollar company?
So apparently if someone knows / guesses the name of your S3 bucket - even if it's private (!) - they can just bankrupt you by sending infinite PUT requests and there is nothing you can do about it.
> requests get rejected
> but AWS still counts it as a write operation against
StackOverflow is laying off 28% of its workforce.
This may be the first large layoff directly due to AI:
> people asking ChatGPT instead of StackOverflow
> usage & ad revenue declines
> having to lay people off to stay profitable / survive
@scalpinjimmy
so many tech platforms seem like only a shadow of their former self. even facebook had something very special like 10 years ago that doesn't really exist anywhere now
The most effective software engineers I know always had a tendency to revert to writing "C" style code.
Nothing seems to come close in terms of readability / maintainability.
Today we're excited to introduce Devin, the first AI software engineer.
Devin is the new state-of-the-art on the SWE-Bench coding benchmark, has successfully passed practical engineering interviews from leading AI companies, and has even completed real jobs on Upwork.
Devin is
the state of Python 2023:
a guy I know working in ML / AI formats his hard drive every time he runs into trouble with python versions / package dependencies
The more I use Llama 3 the more I think that Zuck may have just killed OpenAI and all other large proprietary AI vendors. The gap between latest GPT4 and Llama 70b is virtually non existent. Even if OpenAI releases GPT5 now, 400b Llama 3 is still training and will most likely be
Was GPT4 just lobotomized?
It responds to queries a lot faster but seems to perform a lot worse than just a few weeks ago (not following instructions properly, making very obvious coding mistakes etc)
Quite likely they replaced it with a distilled smaller model to save costs?
An undervalued aspect of LLMs is that you could download a 70b model to your laptop / hard-drive and go to the most remote parts of the world (even space!) and have the majority of human knowledge in an easy to access format (just ask it) - without needing any internet connection
@dmvaldman
might work short term, but I think sooner or later people will catch on to this. The Amazon of 2017 was a completely different site it feels
My first thought when seeing this was that it looks like a scene from a video game.
Then it clicked – they probably trained 90% on synthetically created scenes rendered with Unreal Engine 5.
This would allow near infinite training data you generate procedurally with near real
Introducing Sora, our text-to-video model.
Sora can create videos of up to 60 seconds featuring highly detailed scenes, complex camera motion, and multiple characters with vibrant emotions.
Prompt: “Beautiful, snowy
Thought of starting a little video series around sharing some tech / coding stuff, mainly around browsers - so javascript, css and some of the newer Web APIs, since this is the stack we use for our start up.
Hope you enjoy this first video (excuse my shyness!!)
1/ Youtube can be great but it easy to get stuck watching tutorials and substitute that for building yourself.
It’s important to know how you would solve a problem by yourself.
Problem solving is 99% of coding.
@MimaJonez
their fulfillment / warehouse system does seem to be their biggest moat, but it won't save them long term if their core product experience keeps getting worse
2/ Try to build something small everyday. This is a real shortcut to really up your skill level.
Working on projects is also a great way to build your portfolio.
I just saw a Tech Bro wearing a hoodie and sandals to a formal business meeting. When I asked him why, he said it’s because he’s ‘disrupting the dress code industry’
The most important thing is to be consistent. Coding is difficult and progress will not happen overnight.
Don't be afraid to make mistakes and experiment!
Thank you to everyone who brought this article to our attention. We agree that customers should not have to pay for unauthorized requests that they did not initiate. We’ll have more to share on exactly how we’ll help prevent these charges shortly.
#AWS
#S3
How an empty S3
4/ Concepts > Syntax
Concepts for example OOP, functional programming and separating the Render Logic (visual) from the data / business logic.
Syntax can be Googled anytime.
5/ Coding should be the last step.
Break the problem down first , how would you solve the problem as a human? This can give you clues as how to implement in code.
If Elon succeeds in keeping Twitter running with 80% of workers gone, it might serve as a blueprint for other companies to do the same
What would be the consequences if FAANG etc lay off the majority of their workers? Salaries going down? Laid off people launching own startups?
6/ Learn about the hardware.
How do computers work? How code gets translated to CPU instructions, HTTP, GPU parallelism etc. This will help you to understand why the code works the way it does.
Math at school is taught in such a limited way - no real-world connections, just dead-end problems.
I grew up thinking I was terrible at math, until I started teaching myself. I was shocked at how ‘easy’ and enjoyable it was.
Here are some things that helped me the most:
This is one of the most troubling things I've seen in the last few years -- which may soon be part of >90% browsers thanks to Google's monopoly on browser engines.
When (not if) implemented, it will give Google direct control over the free internet - being able to serve you ads
8/ Explore software engineering in separate contexts: Graphics, Audio, CRUD, Games, Machine Learning …
Understanding multiple areas can help you generalise concepts and apply them anywhere.
when someone from high school gets back in touch with you - it's for one of only 2 possible reasons:
(1) they secretly had a crush on you
(2) multi level marketing scheme
7/ Initially avoid using new shiny frameworks until you have a good foundational understanding of how programming languages and platform technologies work
GPT4 was presumably trained for around 90 days using 25k A100 GPUs.
Microsoft and Meta having reportedly bought 150k H100 GPUs each this year, can now train a GPT4 class model in only 7 days from scratch (not even including their existing GPU stockpile).
A good litmus test to check if someone is going to make it™ in the future:
You are given $40,000 - how do you spend it?
(A) down payment for house
(B) a single H100 GPU
@lukesprosser
I know the same all too well. Kind of like you're running a background process on something until the result is "ready". Perhaps a synergy between the conscious and subconscious
Desperately need some suggestions for good music to have in the background when coding / focusing for many hours.
Hard to do the Startup life when listening to the same things on repeat 🫠🎧
today Apple is releasing the iPhone 11 for the 5th time.
only reason people are forced to buy the new ones is because they cripple the old ones with software "updates"
Unpopular Opinion:
If you work in tech / develop software you probably shouldn’t be spending your money on a newer / faster computer. It might be actually doing you more harm than good.
It’s easy to develop software than runs smooth on the latest 64 Core Threadripper / RTX 4090,
This is one of the most troubling things I've seen in the last few years -- which may soon be part of >90% browsers thanks to Google's monopoly on browser engines.
When (not if) implemented, it will give Google direct control over the free internet - being able to serve you ads
I recently discovered a fan edit of the Hobbit movies that cuts out all the pointless / ridiculous parts & stays as close as possible to the book.
The editor worked on it for over 2 years (incl. custom vfx / audio effects / ...)
It's crazy how much has changed since 2019.
Microsoft struggled for 6 months to train a 340M parameter model to catch up with Google BERT.
Now we have open source Llama 3 which is literally 200x the size with a magnitude better performance that anyone in the world can take as
@stolsvik
aliexpress is usually delivered within a few days since they use local warehouses, at least in Europe. For very expensive items where you save a few hundred $ when buying from alibaba directly it's probably worth the wait
The default path for many start-ups:
get into Y-combinator / accelerator. Use this to raise your series A, work up the metrics for series B etc.
Something I wish would be talked about more:
As soon as you accept VC money, it’s no longer your company. You MUST exit. Which means
@mythosmint
I think you just summed up the only solution. Add in a daily prayer that nobody guesses your bucket name and I guess keep complimenting your AWS account manager so they may grace you with waiving the bill if bad luck ever hits you