AMA with Lukas Biewald, CEO of Weights & Biases – November 4

Hi folks, we’re excited to resume our AMA series. We used to do this in our slack community and folks really liked it, so we’re bringing it back!

The guest

Our first AMA will be with our wonderful CEO @l2k2.

Lukas started building ML models in 2002 and worked with Daphne Koller on NLP research in Stanford’s AI Lab. He briefly spent time at Yahoo and then founded Figure Eight (fka CrowdFlower), a human in the loop platform that transforms unstructured text, image, audio, and video data into customized training data. He sold Figure Eight to Appen, worked at OpenAI briefly and then founded Weights & Biases.

He is also the host of our podcast, Gradient Dissent.

Lukas loves building robots, playing the harmonica, magnet fishing, fossils, and training ML models.

He’s looking forward to answering all your questions about building ML models, best practices for managing ML projects, building robots, running startups and more.

How to participate

  • Post your questions as replies to this thread – you can post anytime between now and the start of the AMA.
  • At 10am on Nov 4, Lukas will pop into the community forum and answer your questions for an hour!

Let’s go!


I heard you once say that if you couldn’t work in the machine learning space, that you would have studied physics. I have two questions about that:

1 - Why physics?
2 - Do you feel that they both tap into the same puzzle-solving areas of the brain, or different but equally compelling parts?

1 Like

How has the recent raise changed how you view and run W&B? Do you still find time to build robots while running a billion dollar company?

Follow up on @onlineinference 's question; what kind of physics? Any thoughts on how theoretical physics seems kind of stuck for the past few decades?

What is the most interesting ML paper you’ve had a chance to read/skim recently?

Hi Lavanya,
Thanks for the opportunity. I would like to ask Lukas a couple of questions, firstly what was the biggest challenge he faced while starting his first startup? Was he not challenged in the same way when he started WandB?
And secondly, was there ever some degree of self doubt when he did his startups? If yes then how he overcame it, and if not then what does he think made him confidant about it?

What is your favourite beverage and why is it Chai? :grin:

1 Like

If you could build/buy a robot to complete any task in your life for you, what would you have it do?

Hi Lukas :vulcan_salute:,
Congratulation for the new funding round and thanks in advance for answering my question.

Today you can undoubtedly be considered a successful engineer and entrepreneur, but I’m sure throughout your career you hit the wall, at least once. I’m speaking one of those periods in which you feel hopeless, you try many things, but nothing seems to go in the right direction and the temptation to just give up on that business idea, on that project is very strong.
How did you overcome such a moment?
How does that event define who you are today?
Thank you Lukas. :grin:

I wish you all the best for your current and future projects. :rocket:


Hey there Lukas,

As someone who led this company from a couple of employees to a place where you have dozens of people working with you, how has your daily work routine changed in the last few years? What aspect of your work has gotten easier, and which part got harder?



Hi Lukas,

There are a couple of questions that I’d like to ask:

  1. Who Has influenced you the most In your career and how?
  2. What Advice Would You Give Your Younger Self?


1 Like

What is your regular day like ?
Have you played with recent/trendy DL architectures like MLPMixer/ConvMLP/FacnyNewTransformerVariant?

1 Like

Hi Lukas,

How did you get your first customers at wandb?

I think my interest in physics comes from really wanting to understand deeply how the world works. I love mathematical puzzles and I love looking at data and trying to find patterns. I sometimes struggled with the abstractness of physics in college and I’m not sure I would be great at it, but I feel like humanity has made a lot of discoveries about the fundamental nature of the universe which I wish I understood better :slight_smile: .

I’m not sure that the recent raise has changed much of my thinking about running W&B. We have more access to capital and will probably grow a little faster and take more risks than we would have otherwise, but I think it’s dangerous to deploy too much capital at once or get too excited about investor excitement - our success will come from customers loving our product.

I wish I had more time to build robots - I think it makes me a smarter and happier CEO over the long term. Lately I’ve been building robots mostly from kits because I just have small windows to make them and kits come with everything you need vs having to go on amazon/alibaba/mcmaster over and over for the one thing you didn’t think of.

1 Like

Theoretical physics does appear to be stuck from the outside but I don’t feel like I understand it well enough to have any thoughts. Astrophysics seems exciting - seems like we’ve made a lot of discoveries since I was in high school. But I’m so far from the state of the art knowledge I would honestly just love to get to a smart undergrad level of understanding.

A lot of people might say this but I was super inspired by the Alphafold paper - I’ve watched a lot of friends work on this problem and it’s so cool to see how much progress has been made: Highly accurate protein structure prediction with AlphaFold | Nature

Hi Lukas! Alot of the W&B audience is self-taught. I know you’ve published a YouTube Intro to ML series but are there any other resources you’d recommend for beginners? Especially ones who, say, don’t know much about code?

Hey Lukas!
How do you switch off and gain some separation from W&B outside of work?

1 Like

When I started my first company I felt like an outsider to silicon valley. Some of you might laugh at that since I went to Stanford but it really was a different era when young founder ceos didn’t seem cool and there were hardly any good resources. I struggled to get each customer and raise every round of venture funding at the time (I mean like pitched hundreds of investors and heard nos before someone any yeses and even that was only after I scraped my way to a significant amount of revenue which was maybe even harder).

I still have a fair amount of self doubt - every startup CEO I know does as well once you get to know them :slight_smile:. I’m not sure it’s really possible to erase that feeling, only to acknowledge it and try to keep moving forward.

Oops I meant that to be a reply to the previous question. Chai is great! But I think Chai in India might be a lot better than Chai in the US for some reason because my Indian upstairs neighbor makes a much better Chai than I get in a coffee shop around here - it’s like a totally different experience.

1 Like

I have a fantasy about a robot that could paint all the walls of my house. I think it would be cool to change up the colors once in a while but painting is such a pain.