Part 2: Making Calculated Risks With Jason Gauci

Show Notes

The conversation with Jason Gauci continues. Jason breaks down the day after you’ve succeeded with your initial champion. What comes next? How do you motivate your team? And how to successfully promote your ideas within an engineering organization.

Jason is the host of the Programming Throwdown podcast. 

What is SWIMM UPSTREAM?

The Swimm Upstream Podcast is a collection of conversations about knowledge sharing, code documentation, change management, scaling dev teams and more. Our guests come from all over the tech world, with some really interesting insights, stories, and… coffee hacks. Join Tom Ahi Dror, Co-Founder of Swimm (a Continuous Documentation platform that streamlines onboarding and knowledge sharing within software engineering teams), as he talks with some of the most inspiring engineering and dev team leaders in the industry.

Tom 0:00
They say that change is the only constant in life. On this season of Swimm Upstream, we're breaking down specific instances of change in software organizations when both technical and human aspects were involved.

And now for part two of our episode with Jason Gauci, Director of Software Engineering at Argo AI and host of the Programming Throwdown podcast.

So what was it about the first success that brought on the second one? What was it in the experience that you had with your first champion that got you in the door?

Jason 0:50
Yeah, that's a good question. So I think what happened was that there are champions all over the company, right? Or if you're a startup, there's champions all over the market, right? But it's hard to raise awareness. They're busy people, there's a lot of things that they're focusing on, they're bouncing between shore winds and risky things and so we did a really good job of publicizing what's going on with this one team. We had a group that was dedicated to reinforcement learning and these other technologies inside the company. We were sharing things to the group like, look, we had this win. So there was another team that was right on Facebook's roadmap, but it was doing a very similar problem to the team that we were helping. And so there was another champion on that team who saw what was going on. So you're like there's an arbitrage opportunity here. Right? And so that’s how it spread.

Tom 1:57
It sounds like this is not only like Jason, which just came here, a few months ago - and with these new ideas, and maybe he doesn't know about Facebook, and what's going on here. And I don't know if I should go down this road with this person. Now, it's Jason, and this team that's been working with him. And they've done stuff. And they crossed a few hurdles. So, you know, it gives them a better, you know, feeling of trust that this makes sense. I should invest at least a little time in this. Right?

Jason 2:31
Exactly. Yeah, that's exactly right.

Tom 2:34
Okay, so the second team, it sounds like you had your work cut out for you, on the second team as well. But you had a starting point? And was that enough? After succeeding with the second one? It did it, click the, you know, everything went from there, because this was on the critical path?

Jason 2:50
Yeah, so what we learned is there's a set of problems that this type of technology is really good for, and for our problems, it isn't. And so, it took years, even after that, to really figure out. You know, a set of rules to know that in advance. In the beginning, even after we had had a few successes, you know, our hit rate was still maybe 50/50, at best. What that meant was, there were, you know, at this point - we're a whole team, and there were people on the team, where they kind of felt like their career was a slot machine. Why pull the slot machine, you know, and I got, I got triple bars, because I worked with this team, and it was a big win, this person next to me pulls the same slot machine and gets nothing.

Jason 3:42
In general, it's extremely challenging to be a leader in a research lab, especially one where you're at Applied Research Lab, servicing the rest of the company for this reason, because the risk profile, in the short term, is not very appetizing. And so, especially junior engineers, who are really intimidated by something like that. You know, what I tell folks is, you know, we're all in it for the long haul. And so in the long run we all have successes. You know, if we work hard, but yeah, to your point, once we got this sort of flywheel going, it's still, it's still a matter of, we need to try it out and cross our fingers. And over time, we built tooling, and we built roads understanding these problems to where we can get good signals in advance.

Tom 4:41
It sounds like it doesn't work in all cases, but it sounds like it's much more mainstream now to work that way? Is that what you expected to happen in the end? Was that surprising?

Jason 4:56
I personally felt like this was the right thing to do. I personally felt like this was going to work out. It made a lot of sense to me, and it was just a matter of finding the right way to do it. What I didn't know is that it would work out before I got fired. And so I remember having this discussion with my family like look, you know, I have a personal reputation to maintain here like things are not going well. We're 18 months in and I haven't really added value to the company. You know, I don't feel good, the company doesn't feel good. Maybe I should just hit the eject button on this like start job hunting and stuff like that. You know, I thought about it, what I decided was, you know, I've never quit anything. I mean, I've left things, I've never quit anything. And I thought I'm going to open-source this. So, if I get fired, I'll have the source code, I can keep working on it and keep this idea going. Of course, the company was more than happy to let me open source something that didn't work. And so I open sourced it and kept working on it, and we got it to work. But there were definitely times where, you know, I felt like it's the right idea. But it just for whatever reason, we just aren't going to be able to make it happen.

Tom 6:20
Okay, so, first of all, this is an amazing story. It reminds me of stories I've heard about academia as well, where you have two researchers, both sitting in their labs on a particular subject matter, and one ends up with a Nobel Prize. The other ends up with not a lot to show. But it's also a matter of, like chance and luck, right? If you want the chance to get that Nobel Prize, you need perseverance.

Jason 6:52
Yeah, exactly. And I think that, you know, you have many, many chances. And so, you know, with the grit and the perseverance, you know, it's really your expected value over all of these chances, and that you have a lot of control over with the right sort of work ethic and discipline.

Tom 7:14
So identifying these chances and taking the right risks. Is that what you mean?

Jason 7:19
Yeah, exactly. Similar to the podcast, where you learn your audience and adapt. So in the podcast, we focus on producing really great content, and we always try to improve on that. And for this project, we were focusing on, can we improve the developer experience? And can we ultimately improve their business metrics? So we just kept focusing on that as we went from person to person, as we talked to different customers inside the company. I think that with enough time, even if the core idea hadn't worked out, we would have pivoted and we would have ultimately been successful. People were just working super hard at it, really passionate about it. I think, broadly speaking, we would have found a way. I was really happy that we were able to get the core idea working.

Tom 8:20
Right now you're working in a very different situation, right? At a startup, as opposed to a very big company. If you were speaking with someone like you and the situation that you were in, what tips or takeaways would you give that person? Maybe you'll say, this is going to be very hard, expect that, and there's nothing you can do about it, but maybe you're gonna have tips for them to make that road a bit quicker.

Jason 8:54
To your first point, I ended up so you know, reAgent now, which is our reinforced flooring library, you can go to reAgent.ai. And, still get the source code, even though it works they let us continue to sync the source code between Facebook and GitHub, so you can go and download it. And I took a pretty long leave of absence when my second son was born. So I took paternity leave, and I took vacation. And then Facebook has a sabbatical program. So I took that. And when I came back, the org was just as healthy as ever. It was amazing. And so I thought, this is an opportunity for me to find sort of my personal next challenge. And so yeah, I've been at Argo for a few months now. And we're spinning up like a machine learning center of excellence, where we're going to help all the different machine learning efforts across the company, and provide a bunch of services and infrastructure for those folks.

Tom 9:55
That does resonate - what you describe in Facebook, right? I mean, am I right?

Jason 10:00
Yeah, exactly. Yeah, it's very similar. I think that it's, it's more broad than reinforcement learning, right? There's, there's a lot of different actually, you know, autonomous vehicles touches on almost every different discipline in some way, shape, or form, right. I think that if you look at the way things move throughout the world, and if you look at the description, or the scene, you could treat a lot of that the same way you would treat prose in a book. And so you can use a lot of the Transformers a lot of that technology to kind of capture a certain scene on the machine learning side.

Jason 10:39
And so yeah, it turns out almost every disables, obviously computer vision. But then obviously, you know, machine control theory, you know, that plays a part as well. So, so it's really a combination of a lot of different things together in one place, which I think is really fascinating.

Jason 10:56
One thing to that, that I learned over the years is to really have a good sense of pace. So, there are times when you might not need to be in the office that much, right or, and now that I guess that metaphor doesn't work, there are times where you don't have to work that much. And then there are times when every single minute counts, right? And so there are times when I would go, before COVID, I would go into the office, and Facebook has an arcade, and I would go to the arcade, I would play games, I would eat some barbecue for lunch, you know, I would talk to some people, I would stop by the web speed team and talk to some friends over there now and make it back to the AI building, you know, and then there were days where I would get into the office at 8am. And I'd actually leave at 8am. Like, I would see the cleaning crew like the same cleaning crew like four times. And I would actually - I had a standing desk. I'd stand for 24 hours straight writing code. And I think that, you know, it's really important, actually, both of those dimensions, like both of those extremes are equally important. I think if you feel, you know, I guess guilty or you feel like you're not contributing, when you're relaxing, then you can't do that time when you feel that burst of energy, that momentum, that inspiration, you know, you're not going to be able to put as much into it there as well. So, you know, be confident in yourself. You know, follow the kind of like your core beliefs, your core philosophy and be confident in them. And then pacing, I think is the second most important thing.

Tom 12:44
Sounds like you recruited some important people along the way that helped make this happen. You explain about how you found them, but any advice about how to recruit someone to your cause?

Jason 12:59
I think an important thing is - don't be afraid of being kind of outspoken. And really promoting your ideas and promoting yourself. I was really passionate about this idea. And so even though we didn't have any proof, it worked. You know, I created this group. And the other thing is similar to what I was saying before: if you're on the real vanguard of something, then there's a ton of opportunity for you to claim a lot of land. I mean, I was actually in Fredericksburg, Texas a couple of weeks ago. And it's a place that has an awesome Pioneer Village where you can see the first people to settle in this area. And so you could get land literally for free. I mean, if you're willing to settle and, and defend yourself against the Native Americans that were constantly like attacking like that was really what was going on. But you'd get huge tracts of land for free. And so similarly, when you're really on the vanguard of something, yeah, there wasn't like a reinforcement learning group. There wasn't a big community. And so there was a big hole to fill there. So in the beginning, you can really, you know, create a whole market there and you can get people really excited. And you know, now there's people who join Facebook who go straight into the reinforcement learning group, even though it didn't even exist before. So...

Tom 14:27
They don't know how much struggle there was. Right?

Jason 14:30
Yeah, that's true. I actually have a story along those lines. You know, I met Vince Cerf one time. It just happened. Somebody asked him, Why are you so famous? You know, you invented I and I have to admit, this is way before my time. So I think he invented TCP or HTTP or something like that, or maybe both? I don't know. But they said, Well, you HTTP is just this thing. I mean, anyone could have invented it. And then he said, you know, the hard part wasn't just writing the spec for this. It was going crisscrossing over the whole country, getting people to adopt this thing. Right?

Tom 15:07
Wow that really resonates with his story, right?

Jason 15:09
Yeah, exactly. And so yeah, I guess, you know, as I said, before, being on the Vanguard or something. It's a lot like wasting your time, you know, and only in hindsight you say, “okay, that was a good idea.” The other thing is, you know, there's so many things I worked on that didn't go anywhere, right?

Tom 14:51
Yeah.

Jason 14:52
You know, I think, that person who got the Nobel Prize, even in your analogy, right, the person next to them did and get the Nobel Prize, but both of them probably did 50-60 projects, and only one Nobel Prize came out of that. And so...

Tom 15:07
Yeah.

Jason 15:07
I actually have a, I actually have a blog, threadpool.cc, where I list all of my various projects, most of them failures. And I just kind of, you know, it's a much more casual kind of blog, and I just kind of talked about, okay, here's the steps I went through. I tried to build my own 3d printer - it melted itself, you know, and this is what happened. And it gives people kind of a feel of what it's like to, you know, have a whole bunch of different threads going and being able to sort of see things fail, see things succeed, but try to learn as much as you can from everything.

Tom 15:48
So I think maybe we can sum up by saying that being on the Vanguard can be the absolute best, but also the absolute worst. Would you agree?

Jason 15:59
Yeah, definitely. And if you have a lot of balls in the air, then it's kind of both at the same time. So at any given time, there's something on fire, and there's something that's, you know, hitting a home run.

Tom 16:50
You have a podcast as well, right?

Jason 16:52
That's right. Yes. So we have Programming Throwdown. We've been doing that for maybe like 12 years or so. We interviewed folks from Swimm and we try to interview as many people all across the industry as possible and we'll mix that in with some shows where it's just Patrick, my co-host, and I talking about various topics. We ultimately set out to teach people not how to code or how to be a good engineer or things you can learn online or from books. We wanted people to feel like what it's like to do that job.

There's this movie, Office Space that came out I think the nineties, and it's just, you know, you're in this cubicle by yourself, you have this one-way communication from this boss with the big coffee mug, right. Do the TBS reports. You know, the reality is being a software engineer is a lot like being an artist.

And I look at these things that we've built. Like each of them has their own Sistine Chapel, and they're works of art in there and they're beautiful. And so, and you know, you commission an artist and you trust an artist, you know, you don't tell an artist what to do. And so, you know, I wanted to. You know, share kind of like how that relationship works and how that feels and that creative energy and let people know, especially high school and college folks know what that's like.

And so we've been, we've been doing that ever since, and it's been a lot of fun. It's been amazing to see the podcast grow. For the first few years, it was just, you know, maybe our parents and so, you know, our moms. I honestly couldn't tell you what made it take off. I think we just kept trying to create better and better content.

Every piece of feedback we got, we listened to it and we tried to adapt and over time, it grew to what it is today.

Tom 16:12
Amazing. This was great. Thank you so much for coming on today. That's all the time that we have. And I hope to do this again, either on this podcast or on yours.

Jason 16:25
Cool. Yeah. Thanks so much for having me. It was really, really fun. And yeah, check out Programming Throwdown. If anyone out there has any questions about how to go zero to one. I think the word is being an entrepreneur. You know how to be an entrepreneur inside of a giant, multinational. Definitely feel free to hit me up on Twitter, neuralnets4life on Twitter.

Tom 19:31
Thanks, everyone for tuning in. That's all the time we have for today. To read episode transcripts, check out our past season, suggest an episode, or join our growing community of developers. Head to swimm.io that Swimm with two M's dot IO.