From grad school to startup acquisition: Keynote speaker Sanjit Biswas


From left to right, Sung-Ju Lee (General Chair of MobiCom 2014) and Sanjit Biswas (Meraki/Cisco)

We welcomed Sanjit Biswas as the first keynote speaker at MobiCom 2014. Sanjit shared his story with us: from graduate school to startup acquisition. He covered the transition from graduate school to startup, and particularly the process of a startup, namely Meraki. Sanjit presented a case study of what it was like to start as a graduate student, from a company like Meraki, which scaled up a bit, and was eventually acquired by Cisco in 2012.

Meraki started as the MIT Roofnet Wireless Mesh Networking Project when Sanjit was in graduate school, pursing his PhD. Sanjit says that the MIT Roofnet project came at a time when people were moving away from NS2 simulations to real testbeds. At the time, there was real industry interest in such systems. It took many years of development and research before papers were published, namely at MobiCom and SigComm in 2005, and the Roofnet project gained much visibility. The main exposure of MIT Roofnet to industry, and the first turning point, started with Google soon after.

Google invited the MIT Roofnet collaborators, namely Sanjit and his colleague John Bicket, to give a talk at the Google headquarters in Mountain View. Sanjit jokes that they had pitched that their mesh networking software could fit on a small chipset, demonstrating the concept using an image that they had photoshopped off of Google, in an easy-to-deploy router box. This was in fact a feat that they had not yet accomplished, but Google was nevertheless interested in this concept and made an order for 1000 units of their final MIT Roofnet product. At the time, Sanjit says that the startup was not the top priority and was not something him and his colleague wanted to do as PhD graduate students. However, their first lucky break was that their advisor was on a 3 to 6 month sabbatical, which gave them the flexibility to try to make their company happen. As it turns out, after their initial efforts, Sanjit and his colleague, John Bicket, decided to quit their PhD and make their picture a reality: Roofnet in a box and focusing their energy on making the router easy to deploy.

The first Meraki team assembled in a rented space in Mountain View next to Google, and at the time, they did not have any outside funding. He was working with talented, smart, and hard-working friends, which he mentions was a very fortunate circumstance, especially as they were all very willing to reason about things, and it was easy to change product direction based on what was beneficial to the company. At the end of his talk, Sanjit emphasizes the importance of working with such people, and recommends to those considering the startup path, to recruit the smartest people they know know and have fun! From his personal experience, recruiting such people would make going to work a fun process, even through all the stress, as the environment would be something you would look forward to go into every day, where it does not feel as much like a job.

He then posits from his experience working with his friends that if you are awesome in graduate school, you will be awesome in a startup. He later builds on this comment by relating to the kind of research a graduate student would do: that systems research is similar to a startup, where one should execute lots of small steps in the right direction. While graduate school requires a 10+ year vision, a startup requires a narrower vision over a 5 year horizon.

At some point in the progress of Meraki, Sanjit and co. decided to raise financial capital. The motivation is that they needed to have an operating or financial buffer to make sure that there was an operating cushion, should something wrong happen. They looked for external funding and investments. Google and Google people were the early investors in the company, and Sanjit and co. were able to make 20M by September 2006.

They then looked for Venture Capitalist (VC) companies, the right investors, that had the right long-term mindset. A question at the end of the talk from the audience prompts Sanjit to elaborate on the choice of a VC company and the role the VC company played in the startup. Sanjit replies that in the Silicon valley, there are different VC firms of different tiers that serve different clients. In the choice of a VC firm, Sanjit says that it is important to reference-check the companies, and him and his colleagues did so by talking to people who were former academics that did the transition to startup. They considered the reputations of the VC companies and ended up working with Sequoia Capital, which in fact did not give the best offer. Sequoia Capital offered less money for more equity, but they had a long-term track record.

As for the kind of control the VC company exercised on their decisions, Sanjit says that the steering wheel was fully in their hands.

In 2007, Sanjit and co. started experimenting with different business models, namely (1) connecting the next billion with $1/month access, (2) municipal-scale outdoor wifi, (3) new hardware platforms for solar and apartment deployment, and (4) ad-supported WiFi with local advertising. He later indicates that though the solar-powered platforms only sold 10 units, it is something he is hoping would take off in a changed market at a different time.

The economic downturn in 2008 affected all companies, including Meraki. The availability of their financial buffer, the 20M investments, allowed them to continue operating, though customers froze their purchases.

Eventually, Meraki began to expand to different markets, such as enterprise markets, higher education, retail, and healthcare. Some interesting large customers approached them, such as California Pizza Kitchen and Motel 6, where the latter is their largest wireless deployment. The benefit of their system is that it is scalable – 10 or 1000 access points can be deployed, which opened them up to different markets.

Throughout the inception of their startup, Sanjit mentions that they were not managed. They in fact figured stuff out on their own: how to run and manage a startup. They were able to figure things out from first principles, without any experience. They got their hands dirty and found ways to test their ideas. He later affirms that running a startup had its fair share of stress and scrapes, responsibilities, and hard decisions with their limited experience and information.

Sanjit is later asked by the audience whether he had any regrets or things he would have done differently, especially since he made the hard decision to leave his PhD. He says that it is hard for him to point out to specific regrets. He adds that he does not have deep regrets, but that his company faced many failed products. The solar-powered node they developed for example, only sold 10 units, and he jokes that the solar power units are still in his office as a reminder. He continues that they probably could have saved a year or two of bringing up their company if they knew what they knew now, but that it worked out eventually.

He is then asked, looking ahead, what’s next for him. He states that he is not compelled to go do anything in particular.

One sound bite Sanjit leaves us with is: “Luck is what happens when preparation meets opportunity.” Though Sanjit and co. were faced with many lucky breaks, they were prepared when the opportunity presented itself to them. Sanjit mentions that some of the lucky breaks that they had in their favor are (1) the market dynamics at the time that rewarded simple, easy to use networking products, (2) the presence of great seed population, from talented colleagues to top tier investors, and (3) they had a head start on the product through the Roofnet research project.


Keynote talks and some words with the speakers.

The keynote speakers this year have been specifically selected to have vast experience in productizing research ideas. The purpose, as Sung-Ju Lee says, is to create a closer relationship between MobiCom and the industry. We have four speakers, whose talks are spread out across the three days of the conference. Below we have short Q&A’s with some of the speakers!

JeffGelhlaar1Jeff Gehlhaar is VP of Technology at Qualcomm Research, one of the most successful wireless company, and leads a very innovative team in the company. His talk is titled, “The Future of Mobile Computing.” Here is some Q&A with Jeff himself:

Lara: What is the most interesting technology we will see from Qualcomm Research that you will be discussing in your talk (for example, a technology that will define or influence our experience with mobile devices in the, perhaps, near future, or a technology that pushes what we are capable of doing using our mobile devices)? Can you give an example user application?

Jeff: We are working on a range of technologies to enable Qualcomm’s vision of a Digital 6th sense, where our mobile devices become increasingly cognitive. The newest and most exciting research we are doing right now is in brain inspired computing with neural networks. For a number of years, Qualcomm Research has been engaged in the application of biologically inspired spiking neural networks to perception problems in mobile devices. Now, we have added a focus on applying Deep Learning which are state of the art solutions to a number of perception problems. What makes this so exciting is our focus on bringing these solutions to mobile whereas others believe this is a problem that can only be solved in the cloud. Together, these technologies are going to make our devices increasingly cognitive. An example is something we are calling Cognitive Photography, where your mobile device will understand the scene you are taking a picture of, take it at the right instant to capture the moment, with the correct camera settings, identify people in the picture, and categorize it for you by content so you can later find that perfect picture.

Lara: Briefly, what are some of the biggest challenges (and research directions) towards making the above technology real that you will be covering in your talk?

Jeff: Typically these neural networks take a great deal of memory and performance. Our challenge is to bring state of the art solutions to mobile in a low power, highly efficient way. This will require innovation in the networks themselves and the hardware Qualcomm will bring to our chips to enable these to run efficiently and at low power. Learn more about these and other Qualcomm Research innovative technology projects at

Lara: As a person from industry who never attended MobiCom, why did you decide to attend? What are your expectations for this conference?

Jeff: It is an honor to be invited to MobiCom on its 20th anniversary, to look back a bit at where we have been, but more importantly to look ahead to what mobile computing will become over the next 20 years. Qualcomm understands the importance of research in helping to propel technology forward. Sponsoring and Participating in conferences such as Mobicom allows us to help foster innovation. It is exciting to have a chance to gather with colleagues who will shape the future of mobile computing.

KitColbert2Kit Colbert is CTO of End-User Computing at VMWare, one of the leaders in visualization software that has growing interest in mobility. Kit’s talk is titled, “BYOzzzz: Focusing on the Unsolved Challenges of Mobility, An Industry Perspective.”

Lara: As a person from industry who never attended MobiCom, why did you decide to attend? What are your expectations for this conference?

Kit: Academic research has always been a cornerstone of VMware. In fact, VMware was started from an academic research project around virtualization. So we want to continue that relationship across all areas of VMware, especially EUC (End-User Computing). MobiCom has a great track record of innovation in the mobility space, making it a highlight of our academic focus. And, let’s be honest, the location this year doesn’t hurt. 🙂

Lara: What is the takeout message that you want to leave your audience – the academic community – with?

Kit: I think there’re two big takeaways for attendees, both of which I’ll highlight in my talk. First, standards are essential. Academia was absolutely critical in the establishment of open standards that paved the way for the Internet and now we’re faced with a similar situation with the Internet of Things, where siloing could easily kill any widespread interoperability. We’d really like help from the community on driving standards. Second, there are many large unsolved problems in the mobility space that we in industry would love academia’s help with, especially in the areas of security and mobile context. These are very ripe areas for innovation and have tremendous value for customers.

biswas-headshotSanjit Biswas, who published his research 10 years ago at MobiCom, founded a startup that was successfully acquired, and is a VP/GM of Cisco Cloud Networking Group. Sanjit’s talk is titled, “From grad school, to startup to acquisition.”


kleinrockProfessor Leonard Kleinrock, most famously known as a father of the Internet, is also an entrepreneur. Kleinrock’s talk is titled, “Some of My Simple Results.” I asked Professor Leonard if he could share with us any personal sound bite, and he says:

“Two months before the Internet came to life back in 1969, I expressed my vision as to what it might become (link). One of the elements I articulated was that the Internet would become invisible, as, for example, electric service is invisible to consumers – one does not care about how electricity is provided, but rather we just plug into a socket in the wall and expect reliable continuous service with a simple human interface.

Unfortunately, the Internet has not yet reached that stage of disappearing into the infrastructure. It is far too complicated to use. Instead of simplicity, it has grown in complexity as far as the end user is concerned.

Now that we are approaching an era of the Internet of Things and a mobile, wireless deployment, there is hope that the Internet will be available everywhere while at the same time it could disappear into the infrastructure and provide its services with a clean, quiet, simple, powerful and invisible interface to the end user.”

Stay tuned for more one-on-one Q&A’s with the speakers and insight into their talks!