Why did you apply for an internship at Google and how supportive was your Ph.D. advisor?
Before my internship at Google, my experience was mostly in academia. I was very curious about the challenges that a company with such amounts of data deals with every day. Moreover, I was extremely interested in experiencing the stimulating environment and culture at Google. I was advised by professor Alessandro Panconesi who was very supportive and encouraged me to apply for an internship at Google.
You interned three times at Google. What projects were you focused on?
Each of the three times I interned, I had the opportunity to work on a distinct research problem with different research groups.
During my most recent internship I joined the Google+ group in the Mountain View, California headquarters working with Sunita Verma. We worked on the problem of friend suggestion, which deals with the challenging issue of suggesting to a given user the people he/she may be interested in adding as a friend. This is an important problem for online social networks, as receiving good friend suggestions significantly improves the user experience.
In one of my previous internships, I joined the AdWords team in New York City working with Jon Feldman where I worked on the problem of automatically identifying, for any given advertiser, who their main competitors in the AdWords system are.
During my first internship in Mountain View working with Alon Altman I worked on defining algorithms for detecting potential attacks in the Google+ network.
Could you share more details about the outcomes of your collaboration with teams at Google?
During all my internships I had the opportunity to closely collaborate with researchers in other teams at Google, in particular with the Graph Mining team in Google Research NY led by Vahab Mirrokni, who is also my Google Doctoral Fellowship mentor. A productive collaboration has continued even after the end of my internships. This joint collaboration with researchers at Google and at Sapienza University has also led to a publication awarded with the best paper award at the 2015 ACK SIGKDD Conference on Knowledge Discovery and Data Mining (KDD).
While the three problems addressed in my internships have very different applications and independent interest, perhaps surprisingly, they can all be tackled by using related graph mining techniques. Both Google+ and AdWords datasets can, in fact, be modeled as a very large scale graph (or network).
In this context, one is interested in designing algorithms that can extract the information needed efficiently (the friends of a user, the competitors of a company, the potential spam users, etc.) while working at Google scale. In all of my internships I also had the opportunity to implement and test these algorithms in the powerful MapReduce infrastructure available at Google on extremely large datasets with billions of entities.
The approach at Google in evaluating the results of projects is very academic in the sense that rigorous empirical evaluations are conducted to show that the approach proposed actually improves over the state-of-the-art. I was also able to share some results of my work with the public through academic publications.
Did you publish at Google during your internship?
Yes, we successfully published a paper at the 2014 International World Wide Web Conference (WWW) as a result of my internship on the AdWords team in NYC. Moreover, we are currently working on a paper submission based on research done during my last summer internship. We also submitted two patents applications for the algorithms developed during my first two internships.
How closely connected was the work you did during your internships to your Ph.D. topic?
My Ph.D. topic, graph mining, is closely connected with all three of my internships at Google. During my Ph.D. studies, I improved my understanding of several topics in large-scale graph mining, which turned out to be very relevant for addressing important issues at Google, as evidenced by the internships projects I have completed. Among the various techniques that I learned during my Ph.D., graph clustering algorithms and random walks methods have been central to my internships, giving me the chance to use them in concrete scenarios at Google. Moreover, the fact that the paper published during my internship at Google is also part of my Ph.D. dissertation shows the relevance of such research projects to my Ph.D. studies.
What impact has this internship experience had on your Ph.D.?
Besides contributing to my Ph.D. thesis with a publication, the most important impact are the relationships I built with Google researchers. Even after the end of my Ph.D., I am still in close collaboration with various researchers at Google to complete publications stemming from my internships and other research projects. In addition, programming in a professional environment at Google has definitely improved my software engineering skills.
Has this internship experience impacted the way you think about your future career?
Thanks to these internships, I have a clearer understanding of research outside of academia and of software engineering. Before joining Google, I had only experienced research at university and my career focus was limited to academic research. Now I know that conducting research at a company in the industry can be a very relevant career path to consider after obtaining a Ph.D.
Now that you just graduated, what’s next?
I moved to the US to start a postdoc position at Brown University with supervisor Professor Eli Upfal. Our team is currently working in research areas closely related to my Ph.D. studies. I am focusing on algorithmic problems and machine learning methods in the analysis of large-scale datasets with potential applications ranging from social networks to computational biology.
Looking back on your experiences now, why should a Ph.D. student apply for an internship at Google? Do you have any advice to offer?
An internship at Google provides a great opportunity to apply your research skills to very challenging and concrete problems that can be tackled only with the scale of data and resources available at Google. Getting hands-on industry experience with a Google internship can be an inspiration for future academic research, as one gets a glimpse into which research problems are more likely to have a strong impact in practice. Furthermore, taking advantage of all the opportunities offered during a Google internship can boost your Ph.D. studies, by leading to new publications in top conferences. More importantly the internship provides valuable connections with high profile researchers and engineers working at Google, which can have a long-lasting positive impact on one’s career -- regardless of whether you pursue a career in the industry or in academia.
My suggestion is just to apply! Internships are a great way to experience research from a different and fascinating perspective.
Posted by Ariana Palombo, Online Hiring & Insights Team
