jeffshek

His willingness to push against determination, setbacks and overcome obstacles largely shaped who I am today. I’m not close to having his work ethic and tenacity, but it is something I aim for.

RIP 24.

I'm not even certain of how that could be objectively measured for children, let alone pets.

A really good podcast about how her strategy on efficient fundraising.

However, I wonder if this changes now with social media - for better or worse, accusations (true and false) are quite quickly thrown.

No-code is used mostly by non-technical founders who do seem quite motivated (enough to do it themselves).

Engineers are more likely to build it themselves (realistically, over-engineer).

It’s in ones best interest to look like they have a close competitor, but in reality so distant it won’t catch up.

Leaderboards are dopamine clickbait for trying to improve.

Why do billionaires want to richer?

Many software engineers (including myself) much prefer the IC route.

It’s hard to see how the incumbents could be beaten — precisely because how effective they are around data, buying potential competitors (Instagram, YouTube) .... but this is precisely because we don’t know what/if the next market shift is.

What happens if AI takes off? What happens if 3D printing magically becomes 100x more efficient and you can print anything you want from home?

We don’t know. It doesn’t seem like the big incumbents could be defeated, but history repeats itself.

To your point though, yes there is business risk of having it on one cloud.

I had a similar impression months ago, and after listening to a few - I was so happy I gave it a fair chance.

This was critical because by then if you weren’t on AWS, you probably didn’t have the DevOps team to do so - MS was now willing to lend you enough consultants to do it for you.

The number of spend shouldn’t be astronomically high. Most workflows in cloud computing are commoditized low-margin. It also explains why Google wants to do high-margin AI stuff, and why AWS is happy with low-margin areas.

Also, a huge amount of cloud costs are based on egress, so having it on one cloud saves you resources.

Even when you think all signs point to a conclusion - think “peak oil hypothesis” something can come out of the moon that changes all your hypothesis, ruining your funds return easily.

Sure, a lot of academics may be embracing PyTorch, but almost all production models have been in TensorFlow. Tesla is a huge notable example that’s using PyTorch at scale.

I do suspect that the split of TensorFlow 1 and 2 is perhaps one of the worst times for TF 2, many teams will likely try out PyTorch instead.

I think both are amazing frameworks, however TF was designed for Google Scale .... which leads to a lot of difficulties since 99.9 are not at Google scale.

"If you want to build a ship, don't drum up people to collect wood and don't assign them tasks and work, but rather teach them to long for the endless immensity of the sea." Antoine de Saint-Exupery

The Mom Test (Book) is a way to verify the thing you're building people will actually want. Providing market-fit can be hard.

Startup School by YC is another good resource to keep you accountable and make progress with your ideas.

VGG19 measure of perceptual loss has been useful for me.