Gunjan works on Bing’s Segment Relevance team. This team focuses on ranking relevance and other machine learning problems for various Bing verticals such as commerce, images, and video. Besides the verticals, they also affect the answers triggering on the main Bing page when the user types segment-specific queries, and have recently also launched improvements to the main Bing engine for commerce queries themselves.
Gunjan’s interest in science in general and computer science in particular started very early (around 7 or 8). Growing up, he was heavily influenced by Carl Sagan, Arthur C. Clarke, and Isaac Asimov’s ideas. Gunjan’s dad (now retired) was an Electrical Engineer, and was also an important influence in getting him excited about science and technology, as he would always answer his questions in great technical detail. By the time Gunjan was in 4th grade he knew he wanted to either build artificial brains, or go into Astronomy or Rocketry. After Gunjan got through IIT-JEE (an engineering entrance test after high-school) he had a choice to go into either of two. Astronomy was also booming because of lots of new discoveries going on at that time. Gunjan’s mentors convinced him to choose computer science over astronomy. Gunjan did not have to regret that he did that; after his undergraduate in India, getting a Ph.D. in Machine Learning from the University of Texas at Austin was fun. To this day, Gunjan maintains a healthy interest in astronomy, and even has a blog for the same (see http:beyondearth.blogspot.com). However, Gunjan’s passion is now in building the most automated, powerful, self-learning, self-upgrading machine learning systems.
Gunjan said he loves his work a lot because its focus is driven by data and user behavior. Often, nothing is what it appears to be, superficially. Gunjan said that instead of a top-down approach where we dream and design a sophisticated app, our users and data tells us what is important and what to predict and exploit next, in order to drive value for our users. The surprise-factor that comes with mining and learning from data is very exciting. Machine Learning and AI in general are coming to age now, as computing power (think map-reduce, distributed computing), software techniques (modern software libraries for everything), and richness of data sources and user interactions (think Kinect, Bing, social-networks, powerful databases, distributed indexes, wisdom of the masses) catch up to a point where you can build very powerful new applications from existing data, that drive huge value and fun to the consumers fairly quickly. Gunjan continued on by saying that many such AI technologies are affecting areas that we could not have imagined before, such as his other area of interest, Astronomy and Space Exploration. Adaptive optics and computerized image recognition and signal processing has allowed thousands of earth-based amateur and professional telescopes to discover hundreds of planets outside our solar-systems, while AI systems helped Deep Space 1 spacecraft to navigate space and self-correct itself several times when connectivity to human operators was several minutes away due to the time it took for the signals to reach Earth. Improvements in automated navigation have allowed the famous Opportunity rover on Mars to drive faster than ever in 2011, after having roamed Mars for over seven years now. Kepler, the first major space-based telescope designed to find extra-terrestrial planets, thanks to improving data-mining and automated navigation and operations capabilities, found thousands of extra-solar planets just in the first six months of operations last year, including dozens in habitable zone that are Earth-like, and could have forests and oceans on them.
Gunjan said that he loves the culture of collaboration of his group, where people work together as a team to deliver complex projects in a short period of time. More specifically in such collaborations, people work on what they are best at, and are able to advise each other a lot. At Bing, senior developers often work on several such projects in parallel, contributing what they are best at. This allow them to have high overall productivity and gives one good job-satisfaction. Of course, Gunjan added, the excellent employee benefits are also an attraction when you have a family. You don’t have to worry about that at Microsoft.
Gunjan said that he and his wife absolutely love living in Seattle. Surprisingly, he said, they like the not-so-hot climate a lot, though he doesn’t mind the long days and extra sun in the summer :-). Coming from Normandy, France, his wife finds the cool climate very good for her. Gunjan said that he loves the fact that in the Seattle area you can get to a desert or to an ocean beach in under 2 and 3 hours respectively, while we are surrounded by unparalleled greenery, beautiful lakes, mountains and volcanoes- all within short driving distances. This coupled with the excellent local food in Seattle area, localized shopping and living, and a short commute for most people, as compared to Bay Area or even Dallas, Texas, is an awesome combination. Also, Gunjan’s 9 year old son really enjoys his Somerset school a lot, which is only 12 minutes from his work location. The schools in Bellevue are great which was one thing he and his wife were looking for when they were trying to find a city to settle down in.