Jump to Navigation

Allen School News

Allen School researchers reveal how smart devices can be turned into surveillance devices with music

CovertBand demo

Researchers from the Allen School’s Networks & Mobile Systems Lab and Security and Privacy Research Lab teamed up on a new project, CovertBand, to demonstrate how smart devices can be converted into surveillance tools capable of secretly tracking the body movements and activities of users and their companions. CovertBand turns off-the-shelf devices into active sonar systems with the help of acoustic pulses concealed in music. The team’s findings reveal how increasingly popular smart home assistants and other connected devices could be used to compromise users’ privacy in their own homes — even from half a world away.

“Most of today’s smart devices including smart TVs, Google Home, Amazon Echo and smartphones come with built-in microphones and speaker systems — which lets us use them to play music, record video and audio tracks, have phone conversations or participate in videoconferencing,” Allen School Ph.D. student and co-lead author Rajalakshmi Nandakumar told UW News. “But that also means that these devices have the basic components in place to make them vulnerable to attack.”

As fellow author and Ph.D. student Alex Takakuwa points out, “Other surveillance approaches require specialized hardware. CovertBand shows for the first time that through-barrier surveillance is possible using no hardware beyond what smart devices already have.”

CovertBand relies on repetitive acoustic pulses in the range of 18 to 20 kHz. While that is typically low enough that most adults are unlikely to pick up on the signals, young people and pets might — and an audible volume is required for more distant surveillance or to pick up activity through walls. To get around this, the team found that they could disguise the pulses under a layer of music, with repetitive, percussive beats the most effective at hiding the additional sound.

Tadayoshi Kohno, Rajalakshmi Nandakumar, Shyam Gollakota

Left to right: Tadayoshi Kohno, Rajalakshmi Nandakumar, and Shyam Gollakota (Not pictured: Alex Takakuwa)

“To our knowledge, this is the first time anyone has demonstrated that it is possible to convert smart commodity devices into active sonar systems using music,” said Allen School professor and co-author Shyam Gollakota.

By connecting a smartphone to a portable speaker or flat-screen TV, the researchers discovered they could use the data collected through CovertBand to accurately identify repetitive movements such as walking, jumping, and exercising up to a distance of six meters within line of sight, and up to three meters through walls. Having proven the concept, researchers believe a combination of more data and the use of machine learning tools would enable rapid classification of a greater variety of movements — and perhaps enable the identification of the individual making them.

With CovertBand, Allen School researchers have identified a plausible threat, given the increasing ubiquity of these devices in our pockets and in our living rooms. But our embrace of emerging technologies needn’t end on a sour note. As professor and co-author Tadayoshi Kohno points out, when it comes to cybersecurity, knowledge is power.

“We’re providing education about what is possible and what capabilities the general public might not know about, so that people can be aware and can build defenses against this,” he said.

The researchers will present a paper detailing their findings at the Ubicomp 2017 conference in Maui, Hawaii next month.

Read the full UW News release here. Learn more and listen to samples of the CovertBand attack music on the project web page here. Check out articles on CovertBand in Fast CompanyDigital Trends, New Atlas, and The Register.

August 16, 2017

Allen School professor Franziska Roesner recognized with TR35 Award

Franziska RoesnerAllen School professor Franziska Roesner has been recognized with a 2017 TR35 Award, MIT Technology Review’s annual celebration of the world’s 35 top innovators under the age of 35. Roesner is honored in the “Inventors” category, recognizing the visionary individuals who are creating the breakthroughs and building the technologies that will shape the future.

Roesner co-directs the Allen School’s Security and Privacy Research Lab, where she analyzes the security and privacy risks of existing and emerging technologies and develops tools to safeguard end users. She is also a member of the University of Washington’s interdisciplinary Tech Policy Lab.

She is the first computer scientist to analyze the risks associated with augmented reality (AR) technologies in order to support the design of systems that mitigate vulnerabilities in these emerging platforms. These technologies are becoming increasingly popular, not only for entertainment but also for assistive purposes, such as heads-up windshield displays in cars. When Roesner began studying them in 2011, products such as Google Glass had not been announced yet and such technologies were still largely in the realm of science fiction. Roesner’s research covers issues associated with both inputs and outputs, from the potentially sensitive sensor data these platforms collect on users in the course of their interactions, to the impact of visual ad content on the safety of users and bystanders. Her impact in AR and virtual reality (VR) extends beyond the lab: her research has made her a go-to source for other researchers, government regulators, and industry leaders on how to counter the privacy, security, and safety risks in order to realize the full potential of these emerging technologies.

Web privacy and security is another area in which Roesner has produced pioneering research that has had a lasting impact on users. In 2011, when web tracking was a nascent concern, she produced the first comprehensive measurement of third-party tracking on the web. More recently, her team studied the evolution of tracking methods over a 20-year period, from 1996 to 2016 using a novel tool called Tracking Excavator. Roesner previously built a new anti-tracking tool, ShareMeNot, whose code was incorporated into the Electronic Frontier Foundation’s PrivacyBadger browser add-on. PrivacyBadger and other add-ons that incorporated ShareMeNot’s ideas are used by millions of people to safeguard their privacy online.

Another user group that has benefitted from Roesner’s user-centric research is journalists and others who rely on secure communication with sources, clients, and colleagues. After hearing stories like how it took reporter Glenn Greenwald months to establish a secure email connection with source Edward Snowden, she collaborated with experts from the journalism community on a study of the computer security needs of journalists and lawyers. Based on those findings, Roesner spearheaded the development of Confidante, a usable encrypted email client that offers the security of traditional encryption technologies without the friction of traditional key management and verification.

“Ideally, we’d like to design and build security and privacy tools that actually work for end users. But to do that, we need to engage with those users, to understand what they need, and not build technology in isolation,” Roesner told UW News.

“As our technologies progress and become even more integral to our lives, the push to consider privacy and security issues will only increase,” she said.

Before joining the UW faculty in 2014, Roesner earned her Ph.D. and Master’s degree from the Allen School working with professor Tadayoshi Kohno, and bachelor’s degrees in computer science and liberal arts from the University of Texas at Austin.

Since 1999, MIT Technology Review has published its annual list of “Innovators Under 35” recognizing exceptional early-career scientists and technologists whose research has the potential to change the world. Past TR35 honorees include Allen School faculty members Shyam Gollakota and Kurtis Heimerl (2014), Jeffrey Heer and Shwetak Patel (2009), and Tadayoshi Kohno (2007), and alumni Kuang Cheng (2014), Noah Snavely (2011), Scott Saponas (2010), Jeffrey Bigham and Adrien Treuille (2009), and Karen Liu and Tapan Parikh (2007).

View Roesner’s TR35 profile here and the full list of 2017 TR35 recipients here.

Congratulations, Franzi!

August 16, 2017

Allen School faculty and alumni gather at annual DARPA ISAT meeting

Every August, the members of the Defense Advanced Research Projects Agency’s Information Science and Technology study group (DARPA ISAT) gather in Woods Hole, Massachusetts for their annual summer meeting to discuss the future of computing and communications technologies.

As professor and past ISAT Chair Ed Lazowska notes, the Allen School tends to be “wonderfully over-represented” at these meetings and on the ISAT study group in general. No fewer than 11 current members have Allen School connections, including three newly appointed to the group this year: professor Jeffrey Heer, director of the Allen School’s Interactive Data Lab, and alumni Ed Felten (Ph.D., ’93), a member of the faculty at Princeton University, and Emin Gun Sirer (Ph.D., ’02), a faculty member at Cornell University.

Since 1987, the U.S. Department of Defense has relied on the 30 scientists and engineers who serve on the DARPA ISAT study group to provide ongoing, independent assessment of the state of advanced information science and technology and to recommend new directions in research.

Allen School faculty and alumni at the 2017 DARPA ISAT study group meeting

The Allen School/DARPA ISAT family. Front row, left to right: bachelor’s alumnus Hakim Weatherspoon (professor at Cornell); professors Ras Bodik, Luis Ceze, and Jeffrey Heer; and Ph.D. alumnus Brandon Lucia (professor at Carnegie Mellon). Back row, left to right: Ph.D. alumna Roxana Geambasu (professor at Columbia); professor Ed Lazowska; adjunct professor Tom Daniel (UW Biology); and Ph.D. alumnus Ed Felten (professor at Princeton). (Not present: professor and Ph.D. alumna Franziska Roesner and Ph.D. alumnus Emin Gun Sirer (professor at Cornell).)

August 15, 2017

Allen School researchers expose cybersecurity risks of DNA sequencing software

Lee Organick, Karl Koscher, and Peter Ney prepare the DNA exploit

Left to right: Lee Organick, Karl Koscher, and Peter Ney prepare the DNA exploit.

In an illustration of just how narrow the divide between the biological and digital worlds has become, a team of researchers from the Allen School released a study revealing potential security risks in software commonly used for DNA sequencing and analysis — and demonstrated for the first time that it is possible to infect software systems with malware delivered via DNA molecules. The team will present its paper, “Computer Security, Privacy, and DNA Sequencing: Compromising Computers with Synthesized DNA, Privacy Leaks, and More,” at the USENIX Security Symposium in Vancouver, British Columbia next week.

Many open-source systems used in DNA analysis began in the cloistered domain of the research lab. As the cost of DNA sequencing has plummeted, new medical and consumer-oriented services have taken advantage, leading to more widespread use — and with it, potential for abuse. While there is no evidence to indicate that DNA sequencing software is at imminent risk, the researchers say now would be a good time to address potential vulnerabilities.

“One of the big things we try to do in the computer security community is to avoid a situation where we say, ‘Oh shoot, adversaries are here and knocking on our door and we’re not prepared,’” said professor Tadayoshi Kohno, co-director of the Security and Privacy Research Lab, in a UW News release.

Tabloid headline: "Computer Virus Spreads to Humans!"

Researcher Tadayoshi Kohno wondered if what this tabloid headline suggested would work in reverse: Could DNA be used to deliver a virus to a computer?

Kohno and Karl Koscher (Ph.D., ’14), who works with Kohno in the Security and Privacy Research Lab, have been down this road before — literally as well as figuratively. In 2010, they and a group of fellow UW and University of California, San Diego security researchers demonstrated that it was possible to hack into modern automobile systems connected to the internet. They have also explored potential security vulnerabilities in implantable medical devices and household robots.

Kohno conceived of this latest experiment after he came across an online discussion about a tabloid headline in which a person was alleged to have been infected by a computer virus. While he wasn’t about to take that fantastical storyline at face value, Kohno was curious whether the concept might work in reverse.

Kohno, Koscher, and Allen School Ph.D. student Peter Ney — representing the cybersecurity side of the equation — teamed up with professor Luis Ceze and research scientist Lee Organick of the Molecular Information Systems Lab, where they are working on an unrelated project to create a DNA-based storage solution for digital data. The group decided not only would they analyze existing software for vulnerabilities; they would attempt to exploit them.

“We wondered whether under semi-realistic circumstances it would be possible to use biological molecules to infect a computer through normal DNA processing,” Ney said.

As it turns out, it is possible. The team introduced a known vulnerability into software they would then use to analyze the DNA sequence. They encoded a malicious exploit within strands of synthetic DNA, and then processed those strands using the compromised software. When they did, the researchers were able to execute the encoded malware to gain control of the computer on which the sample was being analyzed.

While there are a number of physical and technical challenges someone would have to overcome to replicate the experiment in the wild, it nevertheless should serve as a wake-up call for an industry that has not yet had to contend with significant cybersecurity threats. According to Koscher, there are steps companies and labs can immediately take to improve the security of their DNA sequencing software and practice good “security hygiene.”

Onscreen output of DNA sequencing machine,

This output from a DNA sequencing machine includes the team’s exploit.

“There is some really low-hanging fruit out there that people could address just by running standard software analysis tools that will point out security problems and recommend fixes,” he suggested. For the longer term, the group’s recommendations include employing adversarial thinking in setting up new processes, verifying the source of DNA samples prior to processing, and developing the means to detect malicious code in DNA.

The team emphasized that people who use DNA sequencing services should not worry about the security of their personal genetic and medical information — at least, not yet. “Even if someone wanted to do this maliciously, it might not work,” Organick told UW News.

While Ceze admits he is concerned by what the team discovered during their analysis, it is a concern that is largely rooted in conjecture at this point.

“We don’t want to alarm people,” Ceze pointed out. “We do want to give people a heads up that as these molecular and electronic worlds get closer together, there are potential interactions that we haven’t really had to contemplate before.”

Visit the project website and read the UW News release to learn more.

Also see coverage in Wired, The Wall Street Journal, MIT Technology Review, The Atlantic, TechCrunch, Mashable, Gizmodo, ZDNet, GeekWire, Inverse, IEEE Spectrum, and TechRepublic.

August 10, 2017

Allen School’s Zachary Tatlock and Neutrons software verification project featured in Communications of the ACM

Zach Tatlock

Zachary Tatlock

Verifying that software runs safely and reliably is a mandate for mission-critical systems ranging from avionics to automobiles. A system of special significance to professor Zachary Tatlock and colleagues in the Allen School’s Programming Language and System Engineering (PLSE) group is the Clinical Neutron Therapy System (CNTS) at the UW Medical Center. One of only three radioactive therapy systems of its kind in the United States, the CNTS directs powerful radiation to patients’ heads to treat cancers of the tongue and esophagus. As Tatlock notes, “even a small mistake can be potentially deadly.”

In an article featured in the latest issue of Communications of the ACM, Tatlock discusses how he and a team of UW researchers are working to prevent such mistakes through the Neutrons project. The Neutrons project builds on Tatlock’s long-standing interest in devising and improving techniques to ensure that programs remain error free. Verifying systems like the CNTS presents its own set of challenges: the diversity and complexity of the radiotherapy system’s dozen components, each with its own level of criticality, meant that no single tool was suitable for checking critical component properties while also ensuring that their composition implies critical system properties.

Working with Dr. Jonathan Jacky, a radiation oncologist at UW, Tatlock and a team of Allen School researchers that includes professors Emina Torlak, Xi Wang, and Michael Ernst, and graduate students Stuart Pernsteiner and Calvin Loncaric applied modern software verification techniques to construct the first mechanically-checked safety case for a real safety-critical system in clinical use. The safety case includes a detailed formal model of the CNTS and a set of tools for establishing component properties specified by the model. Leveraging existing formal tools, the researchers built the entire case by writing just 2,700 lines of code.

CNTS control panel

CNTS control panel

The CNTS had been operating for over 30 years without incident. However, in constructing the case, Tatlock and the team revealed serious flaws that could potentially result in the beam operating outside of prescribed settings. They also discovered flaws in the implementation of the system’s EPICS language, which they reported to the CNTS staff.

Their work demonstrated that formal, checkable safety cases can provide significant practical benefits by focusing analysis effort on deep properties of system components that matter for the safety of the system as a whole.

“What we want to be able to do is ensure the reliability of all the pieces,” Tatlock explained in the article. “We want to make sure there are no bugs that can affect the parts that are critical.”

The Neutrons project is just one of several undertaken by PLSE researchers that use verification techniques described in the article, including Oeuf, which applies proof assistance to compilers; Verdi, for verifying distributed systems; and Bagpipe, for verifying internet router configurations.

Read the August 2017 ACM article here. Learn more about the Neutrons project in the team’s CAV 2016 paper here, and watch a short video about the project here.

August 8, 2017

UW alumni Vamsi Talla and Justine Sherry win SIGCOMM Doctoral Dissertation Awards

Vamsi Talla and Justine Sherry

Vamsi Talla (left) and Justine Sherry

ACM SIGCOMM has selected two winners of its 2016 Doctoral Dissertation Award recognizing outstanding Ph.D. theses in the fields of computer networking and data communication from the past year, and both recipients have strong ties to the Allen School: former postdoc Vamsi Talla, and bachelor’s alumna Justine Sherry.

Recent Allen School postdoc and UW Electrical Engineering Ph.D. alumnus Vamsi Talla is recognized for “Power, Communication and Sensing Solutions for Energy Constrained Platforms.” Working under the guidance of professors Joshua Smith of the Sensor Systems Lab and Shyam Gollakota of the Networks & Mobile Systems Lab, Talla and his collaborators developed a way to harvest energy from ambient signals such as television and Wi-Fi to power battery-free devices. This technique, known as ambient backscatter, shows great promise for enabling lower-power sensing and communication to realize the full potential of the Internet of Things.

The SIGCOMM award is the latest in a line of honors Talla has received for his work. Previously, he received the ACM SIGMOBILE Doctoral Dissertation Award and the WAGS/UMI Outstanding Innovation in Technology Award. The research team has also won multiple, recent Best Paper Awards for projects related to backscatter, including SIGCOMM 2016, NSDI 2016 and SIGCOMM 2013. Talla is currently the CTO of Jeeva Wireless, a startup company he and his colleagues launched to commercialize their research.

Undergraduate alumna Justine Sherry (B.S., ‘10) is recognized for her Ph.D. dissertation at UC Berkeley titled “Middleboxes as a Cloud Service.” In her thesis, Sherry describes APLOMB, a system that enables enterprises to outsource middlebox processing, such as firewalls and caches, to a third-party cloud service or ISP — providing benefits in terms of scalability and ease of management while reducing costs.

Sherry is now on the faculty at Carnegie Mellon University, where she continues to focus on middleboxes as part of her computer networking research. She previously earned a Best Student Paper Award at SIGCOMM 2015 for presenting a novel design for fault-tolerance in middleboxes. As an undergraduate, she received the Allen School’s Best Senior Thesis Award for her 2015 bachelor’s thesis, “Applications of the IP Timestamp Option to Internet Measurement,” which she completed under the guidance of professors Tom Anderson and Arvind Krishnamurthy. She also co-authored the Best Paper at NSDI 2010 presenting a new internet diagnostic tool, reverse traceroute.

Talla and Sherry will be honored at the SIGCOMM 2017 conference later this month in Los Angeles, California.

Congratulations, Vamsi and Justine!

August 7, 2017

Allen School and Microsoft Research gather leading researchers to envision the future of IoT

Affan Dar

Affan Dar, Principal Group Software Engineering Manager for the Azure IoT Platform at Microsoft, delivers the opening keynote of the 2017 Summer Institute.

A group of leading researchers from academia and industry are gathered in Snoqualmie, Washington this week to explore the future direction of one of the fastest-growing areas of computing innovation: the Internet of Things. Organized by the Allen School and Microsoft Research, the 2017 Summer Institute: Unpacking the Future of IoT aims to foster new ideas, collaborations, and excitement around emerging technologies that will touch every part of our lives.

The Internet of Things is already having a profound impact on how we interact with the physical world around us — and with each other. Recent advances in hardware design, low-power sensing, wireless networking, artificial intelligence, and cloud computing have enabled computers to more easily gather and analyze data and react to changes in the physical world. This, in turn, has given rise to new devices and services that are altering the way we learn, work, and play. The global market research firm IDT estimates that, by the year 2020, the IoT market will reach $1.7 trillion and nearly 24 billion connected devices. That is more than double the current number of internet-connected smartphones, PCs, tablets, cars, and wearable devices combined.

“The Internet of Things is transforming multiple industries as we speak,” said professor Shyam Gollakota, one of the organizers of this year’s institute and director of the Allen School’s Networks & Mobile Systems Lab. “We have brought together some of the best minds in the world who are defining the business and research opportunities and influencing the direction of IoT.”

Rajalakshmi Nandakumar

Allen School Ph.D. student Rajalakshmi Nandakumar, formerly a research assistant at Microsoft Research India, presents her work on interacting with devices using sonar.

The 2017 Institute has brought together these leading minds to discuss a range of topics that will help shape the future of IoT, including battery-free computing, backscatter communication, human-computer interaction, mobile sensing, edge computing, and security and privacy in relation to IoT technologies. By the end of the conference, participants will have a fuller understanding of the state of the art in IoT — and, organizers hope, be inspired to tackle the open research problems that must be addressed to move this exciting field forward. It will take industry and academia working together to realize the full potential of IoT.

“Together, we have worked hard to find a balance between industry and academic research priorities,” said co-organizer Victor Bahl, distinguished scientist and director of mobile and networking research at Microsoft Research in Redmond. “It has been a pleasure partnering with colleagues at UW whose perspective, ideas, and thoughtfulness have helped create an agenda that I am confident will make this an enlightening meeting for all participants.”

In addition to Gollakota and Bahl, the organizing team includes Allen School professors Joshua R. Smith, director of the Sensor Systems Lab; Shwetak Patel, director of the UbiComp Lab; and Tadayoshi Kohno, co-director of the Security and Privacy Research Lab. Participants in this year’s institute, which is by invitation only, include representatives of Amazon, Carnegie Mellon University, Disney Research, Imperial College London, MIT, the National Academy of Sciences, Pioneer Square Labs, Stanford University, and more.

This is the 22nd institute co-sponsored by UW and Microsoft Research. To learn more, visit the 2017 Summer Institute website here and read a related article on the Microsoft Research blog here. Also check out highlights of the conference program on Hamed Haddadi’s blog here.

August 2, 2017

Allen School professor James R. Lee named Simons Investigator

James LeeProfessor James R. Lee, a member of the Allen School’s Theory of Computation group, has been named a Simons Investigator by the Simons Foundation’s Division of Mathematical and Physical Sciences. Lee, whose research interests span algorithms, optimization, computational complexity theory, and related mathematical fields, is one of a small number of mathematicians, physicists, and computer scientists recognized by the Simons Foundation this year.

The Simons Investigator program is designed to support outstanding scientists in their most productive years, when they are establishing creative new research directions, providing leadership to the field, and effectively mentoring junior scientists. Investigators receive $100,000 annually from the Foundation for a period of five to 10 years.

Lee has devoted his career to exploring the mathematical phenomena that underlie optimization problems and related computational processes. He was a co-recipient of the STOC 2015 Best Paper Award for work demonstrating the inherent limitations of semi-definite programs for solving NP-hard optimization problems. In selecting Lee for an Investigator Award, the Simons Foundation also cited his work on spectral algorithms for graph clustering problems and his application of novel tools from geometry and probability to the theory of computation.

Learn more about the 2017 Simons Investigators here.

Congratulations, James!

August 1, 2017

Pooja Sethi: The “badass and brilliant” Allen School alumna who is not afraid to fail

Pooja Sethi in cap and gown by Drumheller FountainThe latest installment of our Undergrad Spotlight features Pooja Sethi, a native of Granite Falls, Washington who graduated from the Allen School with honors last month. Sethi received her bachelor’s degree in computer engineering and was recognized as one of two winners of our Best Senior Thesis Award, which she earned for “Respeak: A Voice-based, Crowd-powered, and Accessible Speech Transcription System” that she worked on with professor Richard Anderson and Ph.D. student Aditya Vashistha in the Allen School’s Information & Communication Technology for Development Lab.

In addition to building a reputation as an outstanding undergraduate researcher, during the course of her studies Sethi participated in five internships, collaborated on research with the Allen Institute of Artificial Intelligence, served as an undergraduate teaching assistant, and served as an officer for the local chapters of ACM and ACM-W.

Sethi plans to spend the summer working with professor Yejin Choi of the Allen School’s Natural Language Processing research group. After that, she will head to Menlo Park for full-time employment with Facebook. The Allen School took an opportunity, before Sethi leaves campus for California, to find out more about what attracted her to computer science, what she looks forward to next, and what advice she would give to students following in her footsteps.

Allen School: Why did you choose to study CSE?

Pooja Sethi: It’s kind of funny, but growing up, I thought the last thing I would ever want to be was a software engineer. The title always conjured up images of people sitting in lonely cubicles staring at blinking lights and yawning. I’ve also always wanted to do something that would help other people, so since I was a kid I thought I would be a doctor.

In high school, I discovered I really loved math and physics and being able to understand the logic behind how the world works. That motivated me to take my first programming class, and I realized how fun it was to be able to “speak” to computers in code. CSE was also really appealing because it has applications to so many facets of life: people here work on everything from finding security vulnerabilities in self-driving cars to using smartphones to detect if a newborn has jaundice. Computers power everything in our lives, and as I was learning to program, it felt empowering to not only understand how computers work but also to be able to create new things with them.

Allen School: What was your favorite CSE course?

PS: That’s so hard — there were so many CSE classes that I loved! I’m going to list a few since I have a hard time updating weights in a priority queue. CSE 332 (Data Structures) was my gateway into fundamental topics and made me feel like a “real” CSE student. My favorite systems courses were CSE 444 (Database Internals) and CSE 452 (Distributed Systems), because let’s be real, it sounds pretty cool to say you’ve implemented a Query Optimizer or a Paxos-based Key Value Service. I also loved CSE 517 (Natural Language Processing) and CSE 573 (Artificial Intelligence), since they introduced me to my current research area of language generation using deep learning. Last but not least, I would highly recommend CSE 599 (Entrepreneurship) with professor Ed Lazowska and Madrona Venture Group’s Greg Gottesman. It was really fun to build a product from the ground-up and hear from amazing entrepreneurs in the Seattle area.

Allen School: What did you like most or what will you miss most about being a CSE undergraduate?

PS: The people in CSE are hands down what I will miss the most. I’ve found that there is something about being trapped in a basement with tea, underflowing snacks, Linux machines, and buggy code that is very conducive to building lifelong friendships. I’m also really thankful to all of the advisers, grad students, and professors I worked with who made CSE feel awe-inspiring but also very welcoming. And of course, the gorgeous view of Lake Washington from the Allen Center balcony will be sorely missed!

Allen School: Are there any people in particular you’d like to mention, who helped shape your experience as a student?

PS: A huge thank you to all of these wonderful people: First, the “Never Stop Dreaming” family of my fellow students: Karishma Mandyam, Aishwarya Mandyam, Naomi Musgrave, Hasanthi Neti, Huong Vo, Grant Timmerman, Dan Radion, Karan Goel, Anna Pendleton, Lilian Liang, Michael Petrochuk, and Bran Hagger. They filled my college experience with so much joy, code, and chocolate milk. To my big sisters in the Bay Area: Vinaya Polamreddi and Arathi Mani, for their constant inspiration, goddess-level AI prowess, and (literal) sunshine in my life.

I also want to thank the amazing grad students and faculty who guided me through research: Aditya Vashistha, Richard Anderson, Gaetano Borriello, Yejin Choi, Ari Holtzman, Max Forbes, Talia Ringer, James Wilcox, and Zach Tatlock. Also, the lovely academic advisers — Raven Alexander, Jenifer Hiigli, and Maggie Ryan — for their kindness and for answering my many questions throughout the years.

And finally, to my mom, dad, and brother Aakash. I would not be anywhere I am today without my family.

Allen School: What is next for you after graduation?

PS: I’m really excited to be joining Facebook full-time as a software engineer in the fall in the Bay Area. During the summer, I’ll be at the Allen School working on conversational question generation in NLP, advised by Yejin. I’ll also be working part-time at a local startup called Cue doing NLP and machine learning consulting. I hope to go back to school for a Ph.D. in the semi-near future. I also plan on sleeping, climbing, and running more post-graduation; we shall see if I succeed.

Allen School: What advice do you have for younger CSE students coming after you?

PS: There is a stigma associated with talking about failures, especially in CSE where many students come in having gotten used to being top of their class and always getting things right. The truth is, there are many times where I did not get things right — I’ve been rejected from internships, received mediocre scores on exams, and questioned my ability to do this subject at three o’clock in the morning.

I’ve learned that what ends up mattering is not the failures themselves, but how you react to them and the perseverance you build by standing back up. So, to those students I say stand strong, remember that you are badass and brilliant (as a few of my friends like to say), and last but not least, stay humble and kind. There is always more to learn.

Allen School: Any final thoughts you’d like to share, as you get ready to start your next chapter?

PS: Thank you for an incredible four years. To new students coming in: you’re in for quite the adventure!


Good luck, Pooja, and thank you for being an exemplary member of the Allen School community — we look forward to seeing you at a future alumni event!

July 28, 2017

Domino effect: UW and Microsoft researchers use spatial organization to build DNA computers

Rendering of nanoscale circuit boardsA team of researchers from the University of Washington and Microsoft Research Cambridge have achieved a breakthrough in the design of computational circuit boards using DNA molecules. The group, which was led by Allen School and Electrical Engineering professor Georg Seelig and Andrew Phillips, head of the Biological Computation Group at Microsoft Research, applied principles of spatial organization to create a new “DNA domino” architecture that reduced computation time from hours to minutes compared to existing approaches. The results of the team’s work, which could lead to important advances in molecular engineering and embedded circuit design, were published in the journal Nature Nanotechnology.

Spatial organization is at the heart of many human-engineered systems, including silicon circuitry. Spatial constraints — the proximity of a system’s components to one another — dictate the flow of information between those components, accelerating it among those close to each other and reducing interference from those that are positioned farther apart. In a departure from this organizing principle, synthetic molecular circuits have focused on chemical specificity — a biological interaction determined by the binding properties of proteins — over proximity to guide the interactions of diffusible components.

The team from UW and Microsoft opted for spatial organization over sequence specificity in devising its DNA domino architecture. As the name suggests, the circuits in this new system transmit information in a cascade, from one adjacent component to the next — akin to a line of dominoes. To construct their circuits, the team laid out short strands of DNA, called staples or hairpins, on long strands of DNA folded into a scaffold known as DNA origami. Using this approach researchers were able to construct elementary Boolean logic gates and transmission lines, as well as a two-input, dual-rail XNOR gate that demonstrated the potential for the domino architecture to serve as a building block for a molecular computer.

Georg Seelig

Georg Seelig

The domino effect resulted in significant improvements in speed and efficiency over current approaches. In one experiment, the domino circuits performed in just seven minutes what would take a circuit built with diffusible components four hours to complete. And because proximity, not specificity, controls the interactions in the domino architecture, components can be duplicated and reused across circuits — enabling more efficient, modular design of molecular control circuits.

Seelig is particularly interested in how this work will translate into novel applications in health care, from new biosensing capabilities to tiny, DNA therapy-delivery robots.

“This work brings moves us closer to being able to use molecular computing for novel therapeutics and diagnostics,” said Seelig. “Not only because computation in a localized circuit is over 10 times as fast as earlier, non-localized circuits, but also because co-localizing functional components on a molecular board will make them easier to deliver to cells.”

In addition to Seelig and Phillips, co-authors of the paper include UW Bioengineering Ph.D. alum Gourab Chatterjee, now a postdoc in the Seelig Quantitative Biology and DNA Nanotechnology Lab; scientist Neil Dalchau of the Biological Computation research group at Microsoft Research Cambridge; and Richard Muscat, former postdoc in the Seelig Lab and now Research Funding Manager at Cancer Research UK.

Learn more in the Nature Nanotechnology paper and related articles on the Microsoft Research Blog and in IEEE Spectrum.

July 26, 2017

« Newer PostsOlder Posts »