Community//

Peter Bailis of Sisu: “You can’t get tenure with just a vague idea”

One thing that Ali told me early on is that one of the best parts of starting a company is building a team that’s not only better than you at each of their respective roles, but can also teach you about them as you build something larger than any of you would individually. Two and […]

The Thrive Global Community welcomes voices from many spheres on our open platform. We publish pieces as written by outside contributors with a wide range of opinions, which don’t necessarily reflect our own. Community stories are not commissioned by our editorial team and must meet our guidelines prior to being published.

One thing that Ali told me early on is that one of the best parts of starting a company is building a team that’s not only better than you at each of their respective roles, but can also teach you about them as you build something larger than any of you would individually. Two and a half years in, I feel like I’ve learned so much from our team here at Sisu, and, from a personal perspective, the lessons I’ve learned across product, engineering, sales, talent, and marketing have been hugely rewarding.

I think that lesson is especially true for us given that our product wouldn’t be possible if you just looked at machine learning, distributed systems, or user experience in a vacuum. The core user experience we have built is enabled by advances in each of these technical fields. And Ali helped me realize that this kind of collaboration probably wouldn’t happen in a vacuum, and likely wouldn’t have happened if we didn’t start it.


As a part of our series called “Meet The Inventors”, I had the pleasure of interviewing Peter Bailis.

Peter Bailis is the founder and CEO of Sisu, the fastest and most comprehensive augmented analytics platform for structured data. Peter is also an assistant professor of Computer Science at Stanford University, where he co-leads Stanford DAWN, a research project focused on making it dramatically easier to build machine learning-enabled applications. He received his Ph.D. from UC Berkeley in 2015, for which he was awarded the ACM SIGMOD Jim Gray Doctoral Dissertation Award, and holds an A.B. from Harvard College in 2011, both in Computer Science.


Thank you so much for doing this with us! Before we dive in, our readers would love to learn a bit more about you. Can you tell us a bit about your “childhood backstory”?

In my first year of graduate school, one of my advisors pointed me to a poem from Herman Melville called “Art” (yes, Moby Dick Melville!) where Melville talks about the tension between audacity and reverence. The poem describes Jacob’s biblical wrestling match with an angel, and the dueling concepts of audacity and reverence struck me as a tangible balance that every creator has to wrestle with.

At the time, my advisor brought up this allegorical wrestling match to stress the importance of both having big ideas in research, but also having the reverence and respect for ideas and literature that’s come before you. In research, you shouldn’t just polish a round ball by making incremental improvements on problems that have already been solved, but you also have to learn from great thinkers that came before you to situate your work, and to build on and synthesize ideas that have already been proven.

Over time, and as I’ve grown both as a researcher and a founder and CEO, I’ve found this lesson to be nearly universally true for creative efforts: to build something great, you have to be audacious enough to think big, and to believe that you can make it happen. But at the same time, understanding the landscape and having reverence for successes before you ensures you’ve got your eyes open and aren’t reinventing the wheel. It’s a tension that ultimately won’t ever resolve but — like art — requires taste to navigate.

Is there a particular book, podcast, or film that made a significant impact on you? Can you share a story or explain why it resonated with you so much?

Growing up, I loved reading Richard Feynman’s stories, including Surely You’re Joking Mr. Feynman! While Feynman had some personal scruples that I’d take issue with today, I was captivated by his enthusiasm for learning, and applying that learning to problems big and small.

In one story, Feynman talks about how as a boy, he took up the practice of repairing radios, which at the time were powered by vacuum tubes. These circuits were complex, but in relative terms are a lot easier to modify than today’s chips, and Feynman taught himself how radios work. He got so good at understanding them (and became known in his local community for doing so), he started to fix them as a way to make money on the side during the Great Depression.

The idea that a little kid could not only teach himself how something as complex as a radio works, but also actually modify the internals, was empowering to me as a kid. Feynman’s sheer enthusiasm for learning and researching rubbed off on me; I felt both a sense of admiration and empathy for Feynman in the joy I found in self-studying math and computer science. Simple stuff, like writing small programs to explore how concepts like fractals and strange attractors worked.

Ok super. Let’s now shift to the main part of our discussion. What was the catalyst that inspired you to invent your product? Can you share the story of your “ah ha” moment with us?

Around 2015, we were in the throes of Big Data, and it was clear that the data storage systems of the future were getting really, really good. I’d just accepted a tenure-track faculty position at Stanford and had seven years to make my name and case for tenure, so I started to think about big, audacious questions upon which I could make a dent.

One of the key questions that kept coming up was: in a world with nearly free storage and nearly infinite computational resources, where’s the bottleneck in the way people use data?

While the data R&D community had spent decades making people’s questions about their data — called queries — run faster, it turned out that, for most people, coming up with the questions in the first place was becoming a real issue.

I kept running into use cases where people would know the metrics they wanted to optimize (e.g., revenue, customer engagement), but they didn’t have the time or energy to dig into their data to understand what was actually impacting them, and what was actionable at a given time (e.g., is revenue up or down for users in California? In Utah? In Idaho? …).

The “ah ha” moment for me was realizing that, taking many of the great ideas from consumer internet at Web 2.0, we could develop similar kinds of ranking and relevance models that could help users determine what queries (and answers) are most relevant to them at any point in time. This would not only help alleviate the bottleneck of asking questions by automating the question-generation process, but it’d help users get a better handle on the massive and growing amount of data that they had. And, by coupling this with new results in “interpretable ML” and statistical testing, we could explain why we made a given recommendation.

There is no shortage of good ideas out there. Many people have good ideas all the time. But people seem to struggle in taking a good idea and translating it into an actual business. How did you overcome this challenge?

There’s no substitute for spending time with users. Even as researchers, we spent weeks working with real people wrangling real datasets, and understanding both what algorithms and what interfaces were most helpful to them. Then we solved the Computer Science challenges involved in scaling them up to massive amounts of data.

Often when people think of a new idea, they dismiss it saying someone else must have thought of it before. How would you recommend that someone go about researching whether or not their idea has already been created?

I think that, provided you are a potential user of your prospective idea, if you don’t have a good solution at hand, then chances are someone hasn’t built a great solution for your needs yet. Most great solutions spread pretty quickly, especially in software, so my sense is that, while you should likely perform some research into alternatives in your space, being first to market is not the hard part. Execution is.

So, provided you believe your realistic Total Addressable Market (TAM) is economically sustainable, I wouldn’t worry much about this problem. In fact, if you look at most really successful products — like Google Search and Slack, Allbirds and Doordash — they are rarely the first products of their kind. They just tend to be sufficiently different and better for their target user to warrant a switch, and then the companies behind them managed to build a great product experience for their users.

Did you have a role model or a person who inspired you to persevere despite the hardships involved in taking the risk of selling a new product?

Ali Ghodsi, one of my graduate school advisors, co-founded Databricks about six years before I founded Sisu, and he’s been a great resource in his perspective around making he leap from academia to Sisu.

One thing that Ali told me early on is that one of the best parts of starting a company is building a team that’s not only better than you at each of their respective roles, but can also teach you about them as you build something larger than any of you would individually. Two and a half years in, I feel like I’ve learned so much from our team here at Sisu, and, from a personal perspective, the lessons I’ve learned across product, engineering, sales, talent, and marketing have been hugely rewarding.

I think that lesson is especially true for us given that our product wouldn’t be possible if you just looked at machine learning, distributed systems, or user experience in a vacuum. The core user experience we have built is enabled by advances in each of these technical fields. And Ali helped me realize that this kind of collaboration probably wouldn’t happen in a vacuum, and likely wouldn’t have happened if we didn’t start it.

For the benefit of our readers, can you share the story, and outline the steps that you went through, from when you thought of the idea, until it finally landed on the store shelves? In particular we’d love to hear about how to file a patent, how to source a good manufacturer, and how to find a retailer to distribute it.

Our version of store shelves is a SaaS analytics platform, but the story follows a similar path. I’ve been working on new interfaces to data and analytics for over half a decade, and it all started with the research I was pursuing in grad school and as a professor at Stanford.

In 2015, I was a twenty-five-year-old who’d just signed up for a seven-year run on the tenure track at Stanford, eager to make my name as a newly-minted assistant professor of computer science. The safe thing would have been to continue to work on my dissertation area of distributed transaction processing.

However, through my research groups, I started to notice something that didn’t quite fit. Within a single data center, the fastest database systems in both research and practice were getting very fast. With a few hundred thousand dollars of hardware, you could execute one read-write transaction per minute for every person on the planet. What kind of user-facing applications needed more scale than that? So, rather than focusing on just making databases faster, I started to dig deeper: where was this data coming from, and was database speed really the bottleneck I wanted to solve?

I found inspiration in a paper from 1971 called “Designing Organizations for an Information-Rich World” by Turing Award winner Herb Simon. As Simon wrote, “In an information-rich world, the wealth of information means a dearth of something else: a scarcity of whatever it is that information consumes. What information consumes is rather obvious: it consumes the attention of its recipients.”

Now, you can’t get tenure with just a vague idea. You have to execute. So I started writing code.

The more I built, the more I realized a few deeper lessons. First, the solution to prioritizing all the data had to come from a different discipline: machine learning and statistics. Second, there was no “technical silver bullet” to help people monitor and diagnose metrics. And third, too many other approaches were looking at these individual problems in isolation — and as a result, they weren’t solving a real problem for users.

At this point, despite early technical and academic success, something kept me up at night. We hadn’t improved our production UX since the first prototypes. We saw companies like Microsoft incorporate our backend into their production systems, but we hadn’t really closed the loop with users. So, I made the tough (but exciting) choice to step away from academia and put together a team of true experts across functions — not just engineering, but also in sales, marketing, product, and design. The best part about this so far is that every team member we’ve added has taught us something new and is helping us ultimately build something bigger than we could have as individuals.

The early stages must have been challenging. Are you able to identify a “tipping point” after making your invention, when you started to see success? Did you start doing anything different? Are there takeaways or lessons that others can learn from that?

Early on, I had the opportunity to spend a few months in Cambridge at MIT working with Sam Madden, a renowned database professor who’d also co-founded a startup called Cambridge Mobile Telematics (CMT) based on his research on mobile driving.

Sam knows databases — his lab built the columnar database prototypes that became Vertica — and yet, at CMT, his team kept running into a simple problem: out of the tens of thousands of different Android phone models that CMT’s software had to support, a handful would consistently cause problems with their software releases. The models behind CMT’s applications were constantly improving, but because each phone handset had slightly different sensors, it was hard to calibrate for every type of phone. Each release was slightly different, so they faced a moving target.

What we learned: while Sam’s team (like many companies today) had the data to figure out what had changed, they didn’t have the time or energy to ask all the questions required to get the answer. Even if they wrote a script to automatically pose all possible questions, given a combinatorial number of possibilities, someone would have to manually pore through all the results. At CMT, people — not databases — were the bottleneck. This was the truly valuable problem we could help people solve.

What are your thoughts about bootstrapping vs looking for venture capital? What is the best way to decide if you should do either one?

I think venture capital is a great route to go if you have conviction about the upside potential for your business, and the upside potential is very long. Especially if you are in an area requiring large capital investment to see R&D payoffs, venture capital may be the only way to raise the capital required to transform your idea into a real product.

We are very blessed that some of the biggest names in Business, VC funding, Sports, and Entertainment read this column. Is there a person in the world, or in the US, with whom you would love to have a private breakfast or lunch, and why? He or she might just see this if we tag them.

I mentioned Herbert Simon earlier. It’s reaching back into the past, but I’d love the chance to talk to Herb about his research and how he would adapt it to today’s society and economy.

Simon defined the concept of Attention Economics, and his research is some of my earliest inspiration for founding Sisu. As Simon wrote 50 years ago, “In an information-rich world, the wealth of information means a dearth of something else: a scarcity of whatever it is that information consumes. What information consumes is rather obvious: it consumes the attention of its recipients.”

Given all of the data generated by modern business as well as the vast amount of information we’re besieged with online everyday, this problem has only gotten worse — and I’d welcome his insight into how we move forward as a society…

Share your comments below. Please read our commenting guidelines before posting. If you have a concern about a comment, report it here.

You might also like...

Community//

Jim Grundy of Sisu Energy: “Stay Calm!”

by Charlie Katz
Community//

Sisu! How to Realize Your Dreams and Become Your Best Self

by Rachel Ellner
Community//

Sisu: Nordic Lifestyle Philosophy that will Transform your Life

by Victoria Lim
We use cookies on our site to give you the best experience possible. By continuing to browse the site, you agree to this use. For more information on how we use cookies, see our Privacy Policy.