It's kind of like you're given secrets about the universe, but it's my job to figure out what those secrets mean.

I had no idea that machine learning have been used to find planets before that.

I've heard of machine learning to find cat faces and YouTube videos and, you know, make names, but never first, like a scientific pursuit.

In 2009 NASA Lodge Kepler Space Telescope and its goal was to find planets.

And it did this by staring up hundreds of thousands of stars.

If something passes in front of one star, you're going to get a decrease in brightness.

And so if you get that dip periodically, that could be a sign that there is a planet there, one strong.

We can't go through all that data successfully.

So Chris and my professor Inter Vanderburgh we're working on a project to make a neural network to help speed up the process.

But in 2013 couplers, the action wheels broke so Kepler couldn't stay steady.

The algorithm that Chris made couldn't work very well on that new scrambled data.

So I was going to modify this neural network so that we can find all the planets that there are in that data set I spent, you know, 40 hours a week working on my laptop.

I went from knowing nothing about machine learning to training my own model using tensorflow.

My code wasn't working exactly right.

For whatever reason, it wasn't learning at all.

I was scared that maybe it was all broken and I didn't find anything.

Knives on Andrew's targets.

Like, do you think these air really?

And I got an email back, like, pretty quickly that said yes.

I think these are great.

Congratulations.

You found your first planet.

I remember screenshot ing it and sending it to my family group chat like, Look, guys, I am finally like I'm a planet hunter.

I found planets like this is the best feeling.

I knew nothing.

At the start of my project, beginner can slowly but surely build up their knowledge and discover things that they never thought were possible.