WHY THIS, WHY NOW
I spent 14 years at Google working on ML for Google Ads and Google Payments. I belabored the idea of leaving for months. Google is the Mecca of machine learning. Even so, when I joined there was such a need for innovation. For an idea to reach real users in production, it used to take like six to nine months. I wanted that to happen in days. By the time I left to start Jarvis, some projects were reaching production in three days. Which was really pretty good and after that, the innovation cycle increased.
But at the end of the day, we had built this big awesome brain, or intelligence engine, but it’s reduced to being in the hands of just a few corporates. But there are many companies that don’t have the resources to hire a bunch of Googlers or to deploy machine learning on their data.
So, we asked, wouldn’t it be great if you can democratize the access to machine learning and then have it applied towards the rest of the enterprises on the planet, not just like the biggest five, ten companies?”
TALKING TECH
Consumers moved online at a very accelerated pace because of Covid, generating a lot of data. And that data is like gold for businesses. But it is not in one place and most companies do not have the in-house talent to build their own intelligence engine and then run it at scale. Most of the small pool of engineers who have expertise both in big data and in machine learning are still concentrated at places like Google.
“Business owners have to figure out their online strategy. But they are struggling. We started Jarvis to give the rest of the enterprises access to the same caliber intelligence engine as the biggest companies have.”
We integrate their data and then Jarvis’s machine learning algorithms generates insights. In terms of being “the gold” for a business, with our algorithms we can predict daily the LTV for every customer. And we can build taste models for every customer and predict with hyper-personalization the three to five products they’re most likely to purchase.
This is something our team of 15 can do today because of things like Google open sourcing TensorFlow, compute getting more accessible, and a general standardization of frameworks. But we’re still in the nascent phase of the machine learning world and there’s still an insane amount of innovation that is happening daily.
LEARNING AND INSPIRATION
At one point, I opened my diary and wrote a list of 42 rules to live by. I started it thinking that if I pass away, I would like my son to have these to read. The first rule is, “If comfortable, change!”
“Another rule I have is that if I don’t see a path of 10x’ing my utility function in the world in 10 years, I need to change. I no longer saw a path to 10x’ing my impact where I was.”
A year ago, I was well on the path of becoming VP at Google with a very comfortable life when I decided to start Jarvis. My Rule #4 is “Always choose the harder path! That’s where maximum growth exists.” Now it’s seems like it was an easy decision to make but I’d be lying if I said that. I wrote in my diary, pages and pages, positives, negatives, things I will miss. All of that. After I told my mom I was going to leave my job to start a company, she would ping me with all these Google benefits, “Google has all of this, why leave?” I was like, mom, I’m a manager of managers. I know these benefits by heart already!”
One of the lessons I took from Google is that it’s all about the people. I thankfully had a bunch of mentors there that I learned this from. It’s a lesson that will carry forward not just in this company but for my whole life. I learned to hyper-optimize for my folks. And now, many of them have joined me here at Jarvis.
“Hire insanely smart people. Then set them free.”
When you hire, my personal philosophy at Google and now at Jarvis is hire insanely smart people. Then set them free. I don’t even like the word “manage.” If anybody requires management, this is wrong company to be at. All of our folks are highly motivated, self-driven, and amazingly talented individuals.
Another lesson that I’ve carried over is about prioritization. At Google you could pick a new challenge every two years. I was lucky to be involved in creating eight $1B+ products in my 14-year tenure. But I also killed six products six to nine months into development, partly because I could not see a path to $1B in revenue. So that’s the type of benchmark we hold ourselves to at Jarvis.
But there are differences. One of our biggest risks is that because we are a bunch of Xooglers here is that we have the tendency to design to be able to handle a hundred million or, God forbid, a billion users showing up on your product on day one. If we give into that kind of thinking – building for millions of users in the first year of a company, that product will never see the light of day. We don’t have the resources to build products at that scale. That was the biggest, most painful thing for all of us to leave behind.
Right now, we can handle 10,000 customers on the same second. It’s not a hundred billion, but I think it’ll be a few more months before we have 10,000 customers in the most optimistic view. So that is good enough for now.
“At a startup you have to adapt to the resources that you have and cannot let perfect be the enemy of good.”
OFF THE CLOCK
My work is fun, and I never have to do anything I don’t like, thankfully. I have achieved that. But in terms of other hobbies, pre-COVID, I used to do a lot of group sports like soccer, cricket, rock climbing. During COVID, I switched to individual sports like rowing. On weekends, I try to spend much of my time with my family, take some motorcycle rides, and am learning to ride a One Wheel, though I am nursing an injury right now from that last activity.
Another hobby I have is painting. As a child, I earned two diplomas in painting in parallel with getting a technical degree. My dad kept asking me, “why are you killing yourself?”
“I think it’s important to always have three new challenges going at any time: one for the right analytical brain, one for the creative brain and one for your body.”
