Thread by @svpino

If you haven't looked into machine learning yet, you better start now.

I started looking seriously into machine learning around spring of 2015.

The field was very different back then.

Just to give you an idea, the top most popular deep learning frameworks didn't exist:

• TensorFlow was released at the end of 2015
• PyTorch in 2016

In just 5 - 6 years we have gone from "read my paper... it's cool" to "holly shit, look what my phone is doing!"

Machine learning has turned the industry upside down.

We have gone from "that's impossible" to "of course we can!" in record time.

I know there's a lot of hype out there. Bullshitters bullshit.

But I don't care about that. I did an inventory of my life and found out that machine learning controls:

• The information I consume
• The things I buy
• The videos I watch
• The games I play

And more.

The hype machine is humming because machine learning is delivering.

Hype has to stay ahead of what's possible. The more we accomplish, the more we'll hype the possibilities.

More hype is a good indication of what's happening.

Let's do an exercise.

Look at everything we have done over the last 5 years, and tell me what do you think will happen in the next 5?

Where's the limit, assuming there's one?

I'm beyond bullish about the future!

There's a problem, though.

The demand for qualified people is out of this world.

I tried to find relevant statistics, but all I got was a 2017 article estimating 300,000 AI practitioners and professionals worldwide.

That seems low. I'll multiply it by 10 just for kicks.

Let's assume there are 3 million AI researchers and practioners worldwide.

In comparison, "the Internet" estimated 24 million worldwide developers in 2019!

That's 8 times more developers... and remember I already multiplied the 2017 by 10!

If you are a software developer, you are probably aware that there's a huge demand for your talent.

Well... there are 24 million-many of you.

Do you see now why companies are willing to pay small fortunes for machine learning engineers?

But there's even more...

A lot of the firepower we have in the field is doing research. This is great, and one of the main reasons we have made a ton of progress.

But this opens a really big gap: who turns research papers into actual work benefiting people?

There are some whacky estimates online about the number of worldwide machine learning engineers that have the skills to implement enterprise-level machine learning solutions.

I got depressed just by looking at the number. I won't repeat it here.

But it is astonishly low.

I'm sure you get the point.

We need people and we need them yesterday.

More importantly, you need to understand that there's a place here for everyone, regardless of your current background, skillset, curriculum, experience, and whatever else you can come up with.

The First Generation had it rough.

They had to come up with the math, write it on paper, build every tool we have today.

That was a tough time. But that time has changed significantly.

I'm not surprised when people recommend a 1-mile long prerequisite list to anyone interested in machine learning.

Maybe you have seen something like this?

Calculus, Linear Algebra, Probabilities & Statistics, Python, R, This and That Theory™, etc.

Sounds familiar?

Assuming *everyone* that wants to contribute has to master the same laundry list is just not real.

All of those prerequisities are valuable and well-intentioned, but they vary a lot depending on your focus.

Some of them are even becoming less relevant every day.

Today we have tools we didn't have 5 - 10 years ago.

These tools abstract a lot of the hard things we had to know before. This is good. This opens the field to more people.

Takeway: I'm 110% percent there's a place for you here.

But what's the rush?

If you are a software developer today, you are already working on cool projects and enjoying the market's demand for your skills.

Why should you consider looking into machine learning?

I've have three reasons, in no particular order.

First, as I hopefully convinced you already, if you think your skills are in demand, wait until you add machine learning to your list.

Augmenting your skills will give you a shitton of optionality.

The second reason was what pushed me towards machine learning.

Many of the problems you'll be solving are hard, scary, unexplored, and with unlimited potential impact.

When machine learning works, it feels magic.

I never got that feeling before.

The third reason is because I think you don't have a choice.

I believe there will be a time when *most* of the software we build will incorporate machine learning in one way or another.

This doesn't mean that "software will all be machine learning."

I remember when "the web will kill desktop software" was the craziest thing ever said.

If your definition of "killing" is "completely irradicating," then no, the web didn't kill desktop software.

But if you get past that technicality... holy shit!

I believe machine learning has a similar role to play.

From how we build software all the way to what we build, I see machine learning becoming the heart of the future.

Today it's kind of cool-only-part-of-big-things.

Tomorrow?

Why would you wait for the future to catch up with you when you have a chance to build that future?

If at this point, anything here made you *remotely* interested, please reach out. I'm happy to answer your questions and help you get started.

This is me → @svpino. Stay tuned.