Travis Oliphant: NumPy, SciPy, Anaconda, Python & Scientific Programming | Lex Fridman Podcast #224

Travis Oliphant is a data scientist, entrepreneur, and creator of NumPy, SciPy, and Anaconda. Please support this podcast by checking out our sponsors:

EPISODE LINKS:

PODCAST INFO:

OUTLINE:
0:00 - Introduction
1:11 - Early programming
22:52 - SciPy
39:46 - Open source
51:29 - NumPy
1:28:44 - Guido van Rossum
1:41:02 - Efficiency
1:49:54 - Objects
1:56:52 - Numba
2:05:58 - Anaconda
2:10:25 - Conda
2:26:01 - Quansight Labs
2:29:37 - OpenTeams
2:37:10 - GitHub
2:42:40 - Marketing
2:47:18 - Great programming
2:58:08 - Hiring
3:02:06 - Advice for young people

SOCIAL:

[lexfridman]

Here are the timestamps. Please check out our sponsors to support this podcast.
0:00 - Introduction & sponsor mentions:
1:11 - Early programming
22:52 - SciPy
39:46 - Open source
51:29 - NumPy
1:28:44 - Guido van Rossum
1:41:02 - Efficiency
1:49:54 - Objects
1:56:52 - Numba
2:05:58 - Anaconda
2:10:25 - Conda
2:26:01 - Quansight Labs
2:29:37 - OpenTeams
2:37:10 - GitHub
2:42:40 - Marketing
2:47:18 - Great programming
2:58:08 - Hiring
3:02:06 - Advice for young people

[achaney]

OMG!!! This is my old boss over at Anaconda. He is one of the most brilliant, yet kind and extremely considerate people I've ever met. Talking to him was like staring into the sun. Even though I felt out of my depth working among so many absolute brilliant people, I tried hard for him. Even his brother, with who I worked with closely was a just an amazing person. The Oliphant's are awesome.

[spencert]

I’ve been using all of these tools for years and had no idea about the story behind their development. This was super interesting and inspiring.

[ItsJustAstronomical]

Wow, Dr. Oliphant taught my undergrad signal processing class. It was by far the most difficult class I've taken in my life and I have a PhD.

[mesokosmos]

Just wow! I didn't expect one single man behind all of these. Immensely impressive.

[Random]

Lex, I’m almost 2 years into my coding career. Your fascination sure does rub off on people! From the bottom of my heart, thank you.

[DavidHarned]

He's 50, created arguably the most useful python data science tools AND kept his hairline.
Something doesn't add up...

[rahulshandilya]

Thank you Travis for Numpy, While doing my Master thesis, I was stuck with compiled Fortran density function theory (DFT) code using shell script. I didnt know how to extend, modify and add extra calculation I needed. Then I came across a python DFT library (gpaw) and within few month with help of Numpy, Scipy I was able to finished it. I started programming in python without learning it at all !! I fell in love with python since then.

[v.j.]

this guy is literally knowledge on every data science topic. Any obstacle within his life is studied, then a solution is implemented. Some people are just a notch above

[noam]

There is a difference between those who hold knowledge and those who hold wisdom. This man holds both. It was a pleasure to hear.

[RayWalker-pythonic]

I got involved in Python programming in 2000. Still at it. Big user of NumPy and SciPy. I didn't know who was behind it until now! Awesome.

[crackyflipside]

Dude, this is perfect timing. I just started learning all of this a few days ago!

[scole]

Love Python and programming. Thought this was Tim Dillon for a second though from the thumbpic.

[johnfunk]

Wow! What a history. I was goofing around with all of those tools while they were coming out but I was in a management job and couldn’t use them enough. Now that I am retired and working for myself, I can use the mature products!!! Thanks Dr Numpy!!!!

[rrb]

Absolutely one of the best podcasts I have heard! Thanks for this and it’s been a very good learning experience for me. Travis was just so amazing!

[alexanderwang]

My father and peers always pressured me to seek some job at some place that makes a lot of money in the programming realm. I loved you talked with someone who recognizes the bigger value of just seeing possibilities and taking them on with the balance of making sure you can secure the needed funding.

[drover]

I honestly believe we are so lucky to be in the era of all these scientists, programmers, physicists and the rest. We are getting content from the OG people that will go down in the history books as legendary. Thank you for everything you do Lex

[NomadicBrian]

Learning Python over the last couple of years has been so much fun. Pandas, Numpy, MatplotLib just blew me away. Even Beautiful Soup HTML scraping and using that to overcome more and more API blocks. Having fun with Python and weekly NFL Indianapolis Colts stats in line graphs and bar charts.

[fuxTaken]

19:00 Thinking in linear algebra terms is super useful! Props to Matlab and other languages that let you program that way.

The output of a fully-connected neural network layer of n nodes can be thought of as an n-dimensional vector (a matrix of dimension 1*n), let's call N. Following a previous layer of m nodes, outputting vector M, we know there are m*n connections in total, each with their own weight which could be arranged into an m*n matrix, let's call W. The entire forward propagation step could simply be thought of as N = φ(M * W), where * is the matrix product and φ is the element-wise activation function.

[ehlergoat]

I could have liked the context of this conversation a hundred times. I love his perspective and breath of experience.
He was so correct concerning compartmentalizing specialist to develop within organizations, and those developers peering into the open source world and want to play. He was so truthful. I need an appreciation button for the many moments of brilliance.