September 26, 2006 11:05 AM PT
Microsoft Research is known as a magnet for the world’s top computer scientists, leaders in their field who are working to advance the state of the art and make an impact on Microsoft’s products and services. But the company’s research labs are also a destination for emerging talent, a place where younger researchers can work with some of the best minds in their field, collaborate with colleagues around the world, and do work that can make an immediate and lasting impact on products that millions of people and businesses rely on every day.
Mentoring in the Real World
Helen Wang, who works in the Systems and Networking group at Microsoft Research Redmond, chose her job over a university position because it offered the best of both worlds—a chance to work on real-world problems in a great research lab, along with an opportunity to mentor graduate students.
“The biggest attraction,” Wang says, “was that Microsoft Research was the best research lab. Its scale, the caliber of its researchers, the freedom it provides offered me the best opportunity to grow and conduct relevant, high-impact research.”
Citing the importance of continuing to work with graduate students, Wang applauds the lab’s internship and post-doc programs, which gives her similar collaboration opportunities to those of a teaching position.
“I’ve worked with nine excellent graduate students so far,” she says, “and I’ve enjoyed the internships as much as the students have.”
In 2003, pursuing her own instincts, she started the Shield project, an effort to create a vulnerability-driven firewall that provides additional defenses against malware by patching network traffic for known software defects—an approach that is easier to test and to deploy than patching defective software on its own. With her colleagues, she also developed a framework that can transform Web pages with embedded scripts into “safe” equivalents for browsers to render. A number of Microsoft product groups already are beginning to make use of these technologies.
“By straddling between academia and industry,” Wang says, “Microsoft Research offers special opportunities that allow researchers to work on pressing, real-world problems.”
Before Kamal Jain joined Microsoft Research as a post-doc researcher in 2000, he visited the lab with his colleague Ramarathnam “Venkie” Venkatesan, who asked him to name three leading computer scientists. He then took Jain to the cafeteria, where he saw two of those researchers having lunch with the Theory group.
“I was attracted to the breadth and depth of talent at Microsoft Research,” Jain says. “Pretty much every area of modern computer and information sciences is represented here, by the leading people in the community.”
Jain, a founding member of the Algorithms, Computation and Ecommerce group, was equally impressed when he had a new service idea that quickly started getting Bill Gates’ input.
“It’s quite amazing,” Jain says, “that we have such easy access to top executives—I could not imagine this happening at another large company.
“I see computation as a microscope to see, study, and control the world and invent at the finest level. We have already seen some of this power in the Internet, biotech, e-commerce, entertainment, and countless other fields,” Jain says. “But computing is still only useful for half the planet—primarily for relatively rich and literate people. Our biggest opportunity is to find ways to make computing more useful for everyone.”
Building and Verification
Aditya Nori, who joined Microsoft Research India in November 2005, is one of the co-founders of the Yogi project, which is building a scalable software-property checker by directly analyzing program executables. By combining software testing with verification, participants in the project believe that their technology can help developers and testers find and fix bugs in their software.
“I came to Microsoft Research,” Nori says, “because it gave me an opportunity to do high-quality work, make an impression on real-world software, and collaborate with some of the best researchers on the planet. The art of building and verifying high-quality software still remains a challenging scientific and engineering problem, and I believe that we will see some exciting advances in this area in the coming years.”
Ralf Herbrich joined Microsoft Research in 2000 after completing his Ph.D. at the Technical University of Berlin. He recently founded the Applied Games group at Microsoft Research Cambridge with Thore Graepel. The group aims to use approximate probabilistic inference to impact both recreational games, as well as the “games” we play when making decisions in the real world.
Although the group is less than a year old, it already has developed technology that has been incorporated into Xbox Live®. After a long night playing the multiplayer beta version of Halo® 2, Herbrich and Graepel began conceiving TrueSkill™, a skill-based ranking system for Xbox Live games that helps gamers find ideal opponents by estimating their relative skills, taking into account the average skills of the gamer, as well as the system’s uncertainty about those skills. It can estimate the probability that two players would “draw” and match them up—helping ensure a challenging game for both.
Herbrich and his team had easy access to Xbox Live data and the guidance of the company’s game studios, and within three months, the Xbox Live team was well on its way to implementing the technology.
“Here at Microsoft Research,” Herbrich says, “I get the chance to put my work right into products—not just in theory, but working directly with the product teams on cool features.”
Strong and Nice
When database researcher Sergey Melnik first visited Microsoft, he expected grandeur.
“I was looking for a sign that said, ‘Next Exit: Microsoft,’ or a couple of Windows®-shaped skyscrapers,” he says. He breezed past the cozy, low-profile campus and found himself lost in a nearby neighborhood. But he eventually found the right building and is working on a metadata management project with his colleague, Phil Bernstein.
Like many other researchers, Melnik came to Microsoft Research to work with some of the best minds around.
“Microsoft Research has some of the strongest—and nicest—database researchers in the world,” he says. “They shaped the field in fundamental ways, and calling them my colleagues is a real privilege.”
Melnik is working with his colleagues to develop general-purpose technologies for managing database schemas and the mappings between them, and they are collaborating with the SQL Server™ team to develop a next-generation data platform for Microsoft .NET. This platform will enable developers to access data stored in relational databases using a higher-level conceptual view, which greatly simplifies the development of data-intensive enterprise applications.
“Experiencing firsthand how industrial-strength software gets forged is fascinating,” Melnik says. “I love writing code, maybe even more than I love writing research papers. And where else can you enjoy it more than at the No. 1 software company?”