Allen School News

Since graduating from the Allen School, Nicki Dell (Ph.D., ‘15) has focused on using technology to “make our computing-mediated world safer and more equitable for everyone.” Her work combines the fields of human-computer interaction (HCI) and computer security and privacy to improve the lives of overlooked communities, specifically those experiencing intimate partner violence (IPV) and home health care workers.

For her contributions, the Allen School recognized Dell with the 2025 Alumni Impact Award, honoring former students with exceptional records of achievement. 

“This award is incredibly meaningful,” Dell said. “Whether developing mobile health tools in low-resource settings or building interventions to protect survivors of intimate partner violence from technology-facilitated harm, I learned at UW that impact isn’t just measured in lines of code or papers published — but in trust earned, dignity upheld and lives made a little better.”

During her time at the Allen School, Dell worked with the late professor Gaetano Borriello and professors Richard Anderson and Linda Shapiro on research that addresses the needs of those in low-resource settings. Growing up in Zimbabwe, she saw how limited resources and poor infrastructure created daily challenges. Dell designed a system that integrates Open Data Kit Scan, a mobile app that digitizes data from paper forms, into the community health worker supply chain in Mozambique. In her dissertation, Dell developed a mobile camera-based system to help improve data collection and disease diagnosis in low-resource environments. After graduating as the Allen School’s 500th Ph.D. student, Dell became a faculty member at the Jacobs Technion-Cornell Institute at Cornell Tech and Cornell University’s Department of Information Science.

“My time at the University of Washington was transformative,” Dell said. “When I began my Ph.D. journey, I never imagined the path it would lead me on — not just through the world of academic research, but into the lives and stories of people who are too often overlooked by the tech industry.”

At Cornell Tech, one line of Dell’s research has focused on mitigating technology-facilitated abuse experienced by survivors of IPV. She found that abusers threaten, harass, intimidate and monitor victims using adversarial authentication techniques to compromise victims’ accounts or devices. Dell and her collaborators also analyzed more than 500 posts in public online forums where potential perpetrators discussed strategies and justification for surveilling their partners. Building off of her research, Dell co-founded the Clinic to End Tech Abuse (CETA). Trained CETA volunteers work directly with survivors of IPV to mitigate any technology-related abuse they are experiencing, such as checking devices for spyware and providing other privacy and safety information and guidance. Her work has informed legislation, including the Safe Connections Act of 2022, upholding survivors’ requests to have themselves or those in their care removed from abusers’ shared phone plans while retaining their phone numbers. Over the years, her work has received eight paper awards and the 2019 Advocate of New York City Award.

Another line of Dell’s research investigates how technology can support home health care workers, who are some of the most under-resourced among the medical workforce. As the director of technological innovation at the Initiative on Home Care Work in the Center for Applied Research on Work (CAROW), she co-leads a multidisciplinary team of scientists, scholars and physicians aiming to improve patient outcomes and working conditions for home health care workers. For example, Dell and her collaborators designed interactive voice assistants, similar to Amazon’s Alexa, that can help home health aides manage day-to-day tasks or give guidance during medical assessments such as monitoring for leg swelling associated with heart failure. She also explored how computer-mediated peer support groups can benefit home health care workers, as well as how technology can help account for all the invisible, or unnoticed, work that these aides do for patients.

In addition to receiving this year’s Alumni Impact Award, Dell was awarded a 2024 MacArthur Foundation Fellowship, also known as the “genius grant.” Her work has also earned her the 2023 SIGCHI Societal Impact Award and a 2018 National Science Foundation CAREER Award.

“I’m immensely grateful for the mentors who challenged and guided me, chiefly among them Gaetano Borriello, whose guidance I carry with me and who is still profoundly missed,” Dell said. “I’m also grateful for the peers who inspired and encouraged me, for the broader Allen School community that cultivated in me both a rigorous technical foundation and a deep sense of purpose and for the communities I’ve had the privilege to work alongside.”

Dell will be formally honored at the 2025 Allen School graduation celebration on June 13. Read more about the Alumni Impact Award.

Read more →

As part of a multiyear initiative, the Allen School reimagined the introduction to programming course series with the goal of better serving the widest range of students across the University of Washington. Since its launch in 2022, CSE 121, 122 and 123, collectively known as the CSE 12X series, have enrolled thousands of students each year, serving as a gateway into computer science and as an essential part of UW’s general education for the entire campus. 

Building a new course series from the ground up was no small feat, but a herculean team effort from many people in the Allen School. The nominated team – lecturer Kasey Champion and professors Elba Garza, Miya Natsuhara, Hunter Schafer and Brett Wortzman – had central roles, including co-teaching the pilot offerings of the courses.

The UW recognized the team behind this transformation with this year’s Distinguished Team Teaching Award as part of the 2025 Awards of Excellence — one of the University’s highest honors, recognizing outstanding alumni, faculty, staff, students and retirees whose achievements support the UW mission. 

“We are thrilled to see our hard work recognized, and I hope we can keep doing what we’re doing: creating a welcoming environment where students of all computing backgrounds and experiences can thrive,” Wortzman said. 

Changing directions

Prior to the CSE 12X series, UW students interested in computer programming would take CSE 142 and CSE 143. These popular courses sparked many students’ interest in computer science, including Natsuhara, who went on to become a teaching assistant (TA) for them. 

But after many years since its introduction in 2004, it was time for a change. 

The previous courses were designed during a time when few students came to the UW with any programming experience. Over the years, however, more students had access to computing classes in high school such as Advanced Placement classes, and the experience level of the students coming into the introductory courses started to vary. At the same time, the field itself had changed. Computer science is a fast growing field, and over two decades, new methods for teaching programming had emerged. 

“It’s important for our curriculum in general to be reviewed every so often to make sure that we’re keeping up with things in terms of content and what kind of teaching is effective,” Wortzman said. “The student population was also changing, and it was time to respond to that.”

The new CSE 12X series design included a number of fundamental changes — starting from the top. Instead of two introductory classes, the material is spread across a sequence of three courses, and students can start at whichever one they feel comfortable in. To help students choose, the team created a guided self-placement tool. The tool is not an exam; it asks students to describe their confidence with various programming topics, test out some sample problems (and see the answers) and then recommends which course would be right for them. 

So far, the course recommendations have been helpful. Across the Winter, Spring and Autumn 2024 quarters, only 3% of students switched from one course in the sequence to another. 

“One of the goals was to offer more points of entry,” said Natsuhara, who received her bachelor’s degree in 2018 and her master’s degree in 2020 both from the Allen School. “It’s created a healthier learning environment, especially for the CSE 121 course. I don’t want students who have never programmed before to feel intimidated that their classmates are talking about the Python apps they’ve coded. Now, those more advanced students can skip ahead in the sequence and everyone is more likely to be around others at the same experience level.”

In the classroom

The team also updated the structure of the courses themselves. Each four-unit course now features two lectures and two sections guided by TAs per week, compared to three lectures in the previous sequence. With this format, each section only covers the previous lecture’s material, so TAs do not have to rush through topics from multiple lectures. Professors introduce new concepts in lectures, and these smaller sections give students the supportive space for practicing problems, which is “where the real learning happens,” Wortzman noted.

In addition, lectures are paired with short pre-class materials that introduce key concepts and provide practice with the new topics. This pre-work, which should take students about 30 minutes to complete, allows professors to use class time for answering questions, running activities and discussing more advanced applications.

By spreading out the content over three quarters, it gave lecturers the time and space to explicitly teach students important skills such as how to debug and fix problems with their code. Previously, students were expected to either know those skills implicitly or learn it on their own time, Natsuhara explained. 

The team revamped the curriculum to reflect not just the variety of student skill levels, but also intended majors. In Autumn 2024, for example, students from more than 70 majors enrolled in one of the CSE 12X courses. Instead of having students learn by solving traditional programming puzzles, the team wanted to appeal to this broad range of student interests by highlighting how programming can be used to solve real-world problems across different disciplines.

One of Natsuhara’s favorite assignments asks students to implement an algorithm that prioritizes patients in an emergency room based on factors such as age, pain level or insurance status. Despite being well-intentioned, the algorithm is intentionally flawed and does not treat patients equitably. Each assignment has a reflection component asking students to grapple with how that algorithm made them feel, how it could be improved and what are some of the unintended consequences. 

Other assignments follow similar veins including implementing algorithms for allocating disaster relief, generating computer-based election forecasts or identifying trends in social media posts.

“We’re helping our students wrestle with the idea that programming and computer science is not an amoral field — there are moral and ethical implications for these things and we should care about them,” Wortzman said.

The team also updated their grading policies to help students master the material. First, instead of starting at 100% and deducting points for inaccuracies like in traditional grading, the courses use mastery grading concepts such as coarse-grained evaluations that assess students on how well they demonstrate their understanding of the key ideas. Second, students are able to learn from their mistakes and resubmit assignments for an updated grade. This system incentivizes students to keep learning and working because “learning does not stop once an assignment is turned in,” Wortzman explained. 

These policies have already helped students find success in their courses. During the 2023 calendar year, only about 1% of students in a CSE 12X course were retaking the class compared to almost 8% in 2018 under the previous CSE 14X series.

After the team built the new introductory series — course structure, curriculum and even grading policies — from the ground up, they expected some pushback or major bumps in the road. The rollout, however, “appeared amazingly smooth,” said Dan Grossman, Allen School professor and vice director. 

“The nominated team members acted as shock absorbers, and the fact that the new course rollout was such a success and felt so smooth is a great testament to what they pulled off and how hard it was,” Grossman said. 

However, none of these efforts would be possible without the help of the TA community and countless others supporting the course instructors and participating in the design. This project was also supported by funding from the Center for Inclusive Computing (CIC) at Northeastern University.

“The task of launching the new courses was enormous, and we could not have done it without the massive army of undergraduate TAs that helped us put it together and keep it running to this day,” Natsuhara said. “We leaned on TAs and gave them more autonomy and responsibilities than we have previously done, and they stepped up in a really big way. We are still making iterations and improvements to the courses, and their continuous work deserves recognition as well.”

Natsuhara is not the only alum on the team; her colleague Schafer earned both his bachelor’s and master’s in computer science from the Allen School in 2016 and 2018, respectively.

Other members of the Allen School community were also nominated for this year’s UW Awards of Excellence. For the Distinguished Staff Award, fiscal specialists Emily Miller and Bree Siegel were nominated for their work in payroll, senior grants manager Stephanie McConnel was nominated as a member of the Collaborative for Research Education (CORE) training team and Vani Mandava, head of engineering for the UW’s Scientific Software Engineering Center in the eScience Institute, was nominated as part of the Post-award Dashboard Team. Director of Information Technology Aaron Timss was also nominated in the Career Achievement category, recognizing individuals for their demonstrated excellence throughout their years of service to the UW.

Read more about the 2025 UW Distinguished Team Teaching Award.

Read more →

Each year, the Association for Computing Machinery (ACM) recognizes the best Ph.D. dissertations in computer science with its ACM Doctoral Dissertation Award. In the 2024 competition, two of the recognized dissertations were the work of Allen School students: award winner Ashish Sharma (Ph.D., ‘24), now a senior applied scientist at Microsoft, and honorable mention recipient Sewon Min (Ph.D., ‘24), a research scientist at the Allen Institute for AI (Ai2) and incoming faculty member at the University of California, Berkeley.

Both Sharma and Min contributed to advances in artificial intelligence, albeit in different domains — highlighting both the variety and quality of AI research in the Allen School.

For his dissertation titled “Human-AI Collaboration to Support Mental Health and Well-Being,” Sharma devised ways to address a fundamental challenge in health care by leveraging AI to make high-quality mental health support available to more people. Meanwhile, in her dissertation titled “Rethinking Data Use in Large Language Models,” Min addressed fundamental challenges in natural language processing (NLP) by developing a new class of language models (LMs) and alternative approaches for how such models are trained. Min also earned the inaugural ACL Doctoral Dissertation Award from the Association for Computational Linguistics for this work.

Ashish Sharma: “Human-AI Collaboration to Support Mental Health and Well-Being”

Therapy can be an effective tool for supporting those with mental health challenges, but barriers such as an ongoing shortage of clinicians, high costs as well as stigma with seeking care can limit access. Instead of fully replacing therapists and clinicians with AI, given the significant risks inherent in this domain, Sharma proposes an alternative: he developed two novel human-AI collaboration systems designed to augment, rather than replace, human providers.

“Augmenting mental health interventions with AI and NLP-based methods has the potential to provide scaffolding that could make quality mental health care accessible to all,” said Sharma,  who completed his dissertation as part of the Allen School’s Behavioral Data Science Group. “By carefully designing human-AI collaboration that is grounded in psychology expertise to truly understand the complexities of mental health, human behavior and user needs, and is rigorously tested for safety and effectiveness, we can empower both those seeking help and those providing it.”

Sharma introduced reinforcement learning-based methods that can understand, measure and give feedback on how empathy is expressed in online peer-to-peer mental health support platforms. While many peer supporters are well-intentioned in helping those who reach out, they may be untrained and unaware of key psychotherapy skills such as empathy that can foster more effective conversations. He then leveraged and evaluated these methods in a randomized trial of 300 real-world peer supporters from TalkLife, one of the largest, global peer support platforms, and found that the AI-based feedback helped peer supporters express empathy more effectively in their conversations. The research received the Best Paper Award at The Web Conference 2021.

Human-AI collaboration can also enhance the accessibility and engagement of self-guided mental health interventions. These “do-it-yourself” methods to learn and practice coping skills are often cognitively demanding and emotionally triggering, making it difficult to implement them on a wider scale, Sharma explained. Building on psychological and cognitive science theories, he developed human-centered NLP methods to help “debug” human thought and support people through the process of cognitive reframing – that is, identifying and overcoming negative thoughts. In a randomized study of more than 15,000 participants, Sharma showed that the system helped participants reframe negative thoughts and informed psychology theory about the processes that lead to positive outcomes. He deployed this system at Mental Health America, which provides mental health tools and resources, and it has been used by over 160,000 users.

“Ashish’s dissertation is highly interdisciplinary and unparalleled in combining fundamental advances in natural language processing with large-scale, positive, immediate impacts on the mental health of large populations,” said Allen School professor Tim Althoff, who advised Sharma. “To date, his research has directly improved mental health services that support more than 10 million people yearly — an exceptional feat for any researcher.”

Prior to the ACM’s recognition of his work, Sharma received one of two William Chan Memorial Dissertation Awards, which are named for the late Allen School graduate student William Chan and recognize dissertations of exceptional merit, as well as a JP Morgan AI Ph.D. Fellowship.

Sewon Min: “Rethinking Data Use in Large Language Models”

Although current LMs including ChatGPT have transformed NLP progress, they still have fundamental issues, such as factuality and privacy, that arise from how they learn to perform new tasks after training. The widespread belief was that LMs obtain new skills on the fly without additional training through in-context learning; however, Min showed that LMs’ in-context learning capabilities are actually based on patterns they learn in their training data, which can be activated in certain ways. Based on this understanding, she introduced a new class of models called nonparametric LMs. 

“This new class of LMs includes learned parameters and a datastore, from which they retrieve information for improved accuracy and updatability,” Min explained. 

During inference, a nonparametric model can identify and reason with relevant text from its datastore, unlike a conventional model that must remember every relevant detail from its training set. Having a datastore present at inference time can help lead to more efficient and flexible LMs. 

“My Ph.D. thesis is about understanding and advancing large language models centered around how they use the very large text corpora they are trained on,” said Min, who was part of the UW NLP group. “My research established the foundations of nonparametric models, and also opened up new avenues for responsible data use, such as enabling data opt-out and credit assignment to data creators.”

These nonparametric LMs include retrieval-augmented generation (RAG), and her research has helped establish the technique. However, “the recent use of RAG has been mainly using an off-the-shelf retrieval model and an off-the-shelf LM and plugging them together without training the model, whereas my research advocates for developing new architectures and training methods that allow for more effective and efficient use of the datastore,” Min explained.

Nonparametric LMs can lead to new approaches to avoid the legal constraints that traditional LMs often run into. It is common practice to train LMs using all available online data, but this approach can lead to concerns with copyrights and crediting data creators. Instead, Min developed a new method based on nonparametric LMs — training LMs using public domain data, while keeping copyrighted or other high-risk data in a datastore that is only accessed during inference and can be modified at any time. 

“Sewon’s thesis identifies bold, impactful and challenging problems that many researchers shy away from, and then designs creative technical solutions to address these problems,” said Allen School professor Hannaneh Hajishirzi, who is also a senior director for NLP research at Ai2 and co-advised Min alongside faculty colleague Luke Zettlemoyer. “Her ambitious, creative and forward-thinking vision is complemented by a foundation of technical, structured, mathematical and analytical strengths, leading to groundbreaking and pioneering research.”

The ACM and ACL honors are not the only accolades Min has earned for her dissertation work; she, too, earned a William Chan Memorial Dissertation Award from the Allen School, as well as the 2024 Western Association of Graduate Schools (WAGS) ProQuest Innovation in Technology Award, which recognizes research that introduces innovative technology as a creative solution to a major problem. During her time at the Allen School, Min received a JP Morgan Ph.D. Fellowship in AI and was also named a 2022 EECS Rising Star.

Read more about the ACM Doctoral Dissertation Awards as well as a related GeekWire story.

Editor’s note: This story was updated on July 30, 2025 to reflect Min’s ACL recognition, which was announced after the original publication date. Read more →

Artificial intelligence existed as both a subfield of computing and a cultural phenomenon long before ChatGPT entered the lexicon in November of 2022. While AI may be decades old, its impact on the way we work, the way we learn and, indeed, the way we live clearly has been accelerating in recent years. What isn’t clear is what comes next; regardless, a growing number of professionals across a range of industries will need the ability to understand, leverage and integrate AI and machine learning as part of their work.

Starting this fall, one option for gaining the necessary knowledge and skills will be the Allen School’s stackable Graduate Certificate in Modern AI Methods, a new part-time evening program designed with the needs of working professionals in mind.

“This new curriculum provides students with the opportunity to gain hands-on experience and build their knowledge of best practices when it comes to widely-used AI and machine learning methods,” said instructor Taylor Kessler Faulkner. “Professionals in a wide range of fields who are interested in applying AI and ML techniques in their work will benefit from this certificate.”

That curriculum comprises four courses taught by Allen School instructors with deep expertise in the field, addressing topics such as deep learning, computer vision and natural language processing and their applications. The series culminates in a final, project-based course that invites students to put what they’ve learned into practice. 

Although the certificate is geared toward working professionals, the Allen School also welcomes applications from recent graduates who want to develop their knowledge and skills in AI. Unlike many other programs of this type, the Graduate Certificate in Modern AI Methods will be delivered in person on the University of Washington’s main campus — which will provide students with multiple benefits beyond the course content.

“Students in the certificate program will have access to UW facilities and face time with faculty and the other students in their cohort,” noted Allen School professor Luke Zettlemoyer, who is also senior research director at Meta FAIR. “It’s a great opportunity for local professionals and recent graduates without a formal education background in computer science to take graduate-level courses in the Allen School.”

Course content will be available only to students enrolled in the program. The courses are designed to be taken sequentially over twelve months, starting in September. 

For those with ambitions of earning a master’s degree, the stackable certificate in Modern AI Methods can be applied towards either of two stacked master’s degree programs currently offered at the UW: the Master of Science in Artificial Intelligence and Machine Learning for Engineering, and the Master’s of Engineering in Multidisciplinary Engineering. More stackable degree options may be added in future.

“AI is having an impact on many professions, both inside and outside of the technology industry, and that impact will continue to grow as new techniques and tools come online,” said Magdalena Balazinska, professor and director of the Allen School and holder of the Bill & Melinda Gates Chair in Computer Science & Engineering. “With the Allen School’s long history of leadership in AI, we embrace our responsibility to help students to acquire the fundamental knowledge and skills that will enable them to leverage the latest advances in their current profession or any new career path they might want to explore.”

While the program is likely to be a good match for individuals with a background in science, technology, engineering or mathematics (STEM) or mathematically-focused business degrees, holders of a bachelor’s in any field with the requisite math and programming skills are welcome to apply. However, for those with a degree in computer science or computer engineering, Kessler Faulkner says, the Allen School’s Professional Master’s Program is likely to be a better fit. Applicants can complete an online self-assessment prior to submitting their application to gauge how well their skills are a match for the certificate program.

The inaugural cohort will start in autumn 2025. The deadline to apply to be part of that cohort is August 1st. Learn more about the stackable Graduate Certificate in Modern AI Methods by visiting the Allen School website and also check out a related story in GeekWire. Read more →

Even the most well-designed and accessible websites may inadvertently have inaccessible elements — advertisements. Pesky pop-ups or bothersome banner ads may be easy for many people to navigate away from, but for those who use screen readers, ads that are not developed with accessibility in mind can make browsing online a frustrating experience. 

Allen School Ph.D. student Christina Yeung alongside professors Franziska Roesner and Tadayoshi Kohno wanted to understand just how problematic inaccessible ads can be to users who rely on screen readers. By auditing how ads use, or do not use, accessible elements and pairing that with interviews with blind participants about their browsing experience, the researchers found that the overall online ad ecosystem is fairly inaccessible for users with screen readers. However, encouraging ad platforms to adhere to existing web accessibility guidelines can help make surfing the web a better experience for everyone. 

The researchers presented their paper “Analyzing the (In)Accessibility of Online Advertisements” at the 2024 ACM Internet Measurement Conference (IMC) in Madrid, Spain, last November where it received the Best Paper Award. 

“Online ads are everywhere and so pervasive. If you’re browsing on your phone, or even have an ad blocker on your laptop — you will still see ads,” lead author Yeung said. “But because ads are designed with the intent to visually tell you what’s going on, for those who are blind and use screen readers, they can be even more problematic in ways that other people might not think about on a day-to-day basis.”

Yeung and her collaborators analyzed the behavior of over 8,000 ads across 90 different websites based on how well they adhere to Web Content Accessibility Guidelines (WCAG) best practices. Over the course of a month, the team looked at whether the ads disclosed their third-party content status to screen readers as well as their use of HTML assistive attributes such as alt-text and aria-labels. These elements ensure that screen readers can perceive images and other non-text elements on the ad. They also tracked the number of interactive elements each ad had and if there was any missing text associated with links or buttons. For an ad with 15 interactive elements, someone who uses the tab key to maneuver through ads would need to press it 15 times to reach other content on the site. If an ad has a button without associated text, instead of telling the user what it does, the screen reader will just say “button.”

The researchers found that the majority of the ads contained inaccessible elements. More than half of the ads had no alt-text at all, or had empty or non-descriptive strings. Many assistive attributes included non-descriptive language such as “ad” or “image.” They also noticed that ad developers were using title attributes to provide information, contrary to WCAG guidelines. Title attributes can provide more context to specific HTML elements, appearing as a tooltip when a user hovers their mouse over the element. However, not all screen readers can consistently interact with them. 

“Inaccessible ads have two primary problems,” Yeung said. “First, people can’t differentiate what the content is, so they can’t even make the decision as to whether or not they want to interact with it. Secondly, ads that are designed poorly really do negatively impact browsing in a way that can be quite cumbersome.”

Yeung then interviewed blind participants who use screen readers to understand just how burdensome these poorly-designed ads can be. All of the participants reported that these ads both distracted and detracted from their web browsing experience as they were difficult to navigate away from. Because many ads did not disclose their third-party status, participants often had to use context clues to identify them. For example, if someone was on a news site and they suddenly hear content about furniture, they would know that the furniture content is the ad. While the researchers did not evaluate pop-up ads in the study, participants brought up how frustrating these ads are because they are difficult to close and participants struggled to get back to where they were on the page before the ad.

Only a few large companies dominate the ad landscape, so refining how they adhere to accessibility guidelines can make a noticeable difference. Major ad platforms such as Google, Yahoo and Criteo could create and enforce policies requiring ads to provide meaningful information to screen readers in the HTML attributes. They could also go a step further and develop templates that encourage using assistive attributes and reject ads with generic or missing information, Yeung explained.

“By making some fairly minor changes, we can improve the ecosystem in a way that makes browsing more equitable for everyone,” Yeung said.

Next, Yeung is looking into people’s perceptions of the data collection practices of different generative artificial intelligence companies. 

Read the full paper on ad inaccessibility.

Read more →

Ever since he was a student at the University of Washington, Armon Dadgar (B.S., ‘11) has had his head in the cloud. And despite co-founding the high-flying company HashiCorp after graduation, he has kept his feet firmly on the ground by finding ways to parlay his success into support for future innovators and entrepreneurs.

That success grew out of an experience Dadgar and his friend and co-founder, Mitchell Hashimoto (B.S., ‘11), had as undergraduate researchers in the Allen School’s systems research group. It was there that the two gained their first hands-on exposure to cloud computing and the challenges it posed for practitioners. At the time, cloud computing was on the rise, and of today’s three big players — Amazon, Google and Microsoft — only Amazon had officially launched its platform. But Dadgar and Hashimoto had access to all three for the aptly named Seattle Project, which aimed to leverage these emerging platforms for large-scale, peer-to-peer scientific applications. As part of the project, the duo attempted to build a software solution that would span the “multi-cloud” environment they had to work with.

They were unsuccessful on that first attempt, but according to Dadgar, the experience sparked their entrepreneurial spirit. After graduation, they moved to San Francisco and eventually decided to revisit the old research problems that had since emerged on an enterprise scale. They started HashiCorp, which became a leading provider of software for companies and organizations seeking to automate their infrastructure and security management in multi-cloud and hybrid environments. As co-founder and Chief Technology Officer, Dadgar helped grow HashiCorp to over 2,500 employees. The company counted such household names as Expedia and Starbucks among its roughly 5,000 commercial customers prior to its acquisition by IBM for $6.4 billion earlier this year, after going public in 2021.

“Major revolutions in computing, such as the public cloud, have depended on crucial research innovation in computer systems. As an undergrad in the Allen School, I was fortunate to have been exposed to research in operating systems, virtualization, networking, and more which underpins the public cloud,” said Dadgar. “Those experiences ultimately led to me founding HashiCorp. By supporting systems research, I hope for the Allen School to continue to be at the forefront of innovation in AI and beyond to inspire the next generation of students, researchers and entrepreneurs.”

Dadgar may have traded the city by the sound for the city by the bay years ago, but his affection for the UW is evergreen. Now, he and his partner, Joshua Kalla, are living in Seattle and hoping to sow the seeds of the next HashiCorp through a $3 million gift to the Allen School to support research and student success — and drive the next wave of systems innovation for the artificial intelligence era. The couple’s commitment includes $1 million to establish the Armon Dadgar & Joshua Kalla Endowed Professorship in Computer Science & Engineering, with the intent to help propel Seattle and Dadgar’s alma mater from the epicenter of cloud computing to the leading edge at the intersection of systems and AI.

“We are incredibly grateful to Armon and Josh for their generosity,” said Magdalena Balazinska, director of the Allen School and Bill & Melinda Gates Chair in Computer Science & Engineering. “The Allen School is one of the top computer science programs in the country, and an academic leader in cloud computing, systems, and AI research. But to maintain that leadership and continue to make transformational advances while educating the next generation of innovators, we need support to attract and retain the most talented faculty and students. Armon’s and Josh’s gift will greatly help us with that.”

While Dadgar is eager to give next-generation systems research a lift, he is even more enthusiastic about elevating the next generation of students entering the field. To that end, he and Kalla have committed $2 million to the Allen School Student Success Fund to support a variety of initiatives aimed at prospective and current Allen School students, with a focus on first-generation college students and K-12 students in Washington with limited access to computing education resources.

“Education has always been an incredible driver of economic mobility,” said Dadgar. “Our goal is to broaden the pathways into computer science and technology, and particularly to focus on first-generation college students where we can have a multi-generational impact on both the individual and their families.”

Dadgar has repeatedly walked the talk, whether on campus or at company headquarters. At HashiCorp, he championed the creation of the Early Career Program in 2021 to enable college students of all majors and backgrounds to spend a summer at HashiCorp applying what they’ve learned in the classroom in a real-world corporate setting. More than 170 interns from across the country have benefited from the program’s mentorship and networking opportunities — over a third of whom accepted full-time positions with the company after graduation. In 2019, Dadgar and Kalla committed $3.6 million to the UW to provide scholarships to undergraduate students who participate in the university’s Educational Opportunity Program, which has supported 35 scholars to date.

As a professor at Yale University, Kalla is well aware of the impact such programs can have on students — and the institutions that provide them with that pathway to economic mobility.

“The University environment is a unique setting where students are exposed to new ideas, learn valuable skills, and through research advance the frontiers of knowledge,” said Kalla. “Creating opportunities for the next generation to participate and ultimately to lead us forward is incredibly important to us personally.”

Among the programs supported by the Student Success Fund are the Allen School Scholars Program, a one-year cohort-based program for incoming computer science and computer engineering majors focused on emerging leaders from first generation, low-income and underserved communities, and Changemakers in Computing, a summer program for rising juniors and seniors in high school to learn about computing and its societal impacts.

“We’re hugely appreciative of Armon and Josh’s extraordinary generosity, which will have a lasting impact on our program and our students,” said Ed Lazowska, professor and the Bill & Melinda Gates Chair Emeritus at the Allen School. “This gift is an opportunity to reflect on the inspirational story of HashiCorp: best friends pursuing a vision that began with some software that they built as part of an undergraduate project in the Allen School — and it will enable future generations of Allen School students to pursue their dreams.”

Read a related GeekWire story.
Read more →

Throughout his career, professor James Fogarty, who joined the Allen School faculty in 2006, has grown to become a central figure in Seattle’s human-computer interaction (HCI) community and beyond. His research has made key contributions in sensor-based interactions, interactive machine learning, personal health informatics and accessibility, publishing over 100 peer-reviewed papers. At the same time, he has played a pivotal role in founding and growing Design, Use, Build (DUB) — University of Washington’s cross-campus HCI alliance bringing together faculty, students, researchers and industry partners.

The ACM Special Interest Group on Computer-Human Interaction (SIGCHI) recognized Fogarty’s contributions and inducted him into the SIGCHI Academy Class of 2025. Each class represents the principal leaders of the field, whose research has helped shape how we think of HCI.

“I am honored to be among the SIGCHI Academy Class of 2025,” Fogarty said. “I’m grateful for the amazing students and collaborators that I’ve had the pleasure to work with over the years, advancing HCI, interactive machine learning, personal health informatics and accessibility research.”

Since the beginning of his career, Fogarty has made breakthroughs in HCI research. As a first-generation student, he was introduced to HCI research at Virginia Tech and was part of the first cohort of Ph.D. students in the Human-Computer Interaction Institute at Carnegie Mellon University. Key research from his dissertation, which focused on using sensor-based interactions to predict the best time to interrupt someone, received a 2005 CHI Best Paper Award. 

Upon joining the UW, Fogarty launched a new research emphasis in interaction with artificial intelligence (AI) and machine learning. Fogarty’s research into new methods for engaging end-users in machine learning training and assessment and understanding difficulties that machine learning developers encounter was considered ahead of its time. The researchers contributed to what is now known as human-AI interaction before it became a trending topic, and this line of research went on to directly impact industry guidelines for the field. 

In the same period, Fogarty and his collaborators developed Prefab, a system for real-time interpretation and enhancement of graphical interfaces through reverse engineering their pixel-level appearance. Prefab, which earned a 2010 CHI Best Paper Award, was a breakthrough in interface systems research, foreshadowing current work using AI to understand, interact with and enhance graphical interfaces. 

Fogarty then turned his research interests toward digital health, including tools to support people in self-tracking and making sense of health data. He and his collaborators provided new insights into the design of tools to support menstrual tracking, informing the design of Apple’s menstrual cycle tracking support. The research received a 2017 CHI Best Paper Award and helped expand the field’s conception of personal informatics to account for the role of design in experiences people have with their tools. Fogarty also researched the design of mobile food journals and activity-tracking visualizations as well as tools to help patients collaborate with their health providers to interpret and act on self-tracked data. More recently, Fogarty and his collaborators developed a new goal-directed approach to long-term migraine tracking, which earned them a 2024 CHI Best Paper Award, and a new tool for home-based self-monitoring of cognitive impairment in patients with liver disease, recognized with a 2025 CHI Best Paper Award. 

Outside of health tracking, Fogarty has also made important strides in accessibility research. He and his team drew inspiration from epidemiology to conduct the first large-scale assessment of accessibility in 10,000 Android apps. The Department of Justice cited the work as part of its updates to the Americans with Disabilities Act. He also extended his work on interface understanding and enhancement to demonstrate real-time repair of mobile app accessibility failures. This research helped directly motivate and inform Apple’s launch of accessibility repair in its pixel-based Screen Recognition.

“James has made an exemplary impact across research disciplines and industry,” Allen School professor Jeffrey Heer said. “His research prowess, volunteer spirit, deep care, thoughtfulness and community-mindedness have helped guide DUB and advance the HCI community in Seattle and across the globe.”

Fogarty is one of five UW faculty being recognized with ACM SIGCHI Awards this year. Department of Human Centered Design & Engineering (HCDE) professor and Allen School adjunct faculty member Kate Starbird joins Fogarty as part of the CHI Academy Class of 2025, and her work sits at the intersection of HCI and computer-supported cooperative work. 

Information School professor and Allen School adjunct faculty member Alexis Hiniker also won an ACM SIGCHI Societal Impact Award for her research into ways that consumer-facing technologies can hurt young people instead of helping them thrive. Nadya Peek and Cecilia Aragon, both HCDE professors and Allen School adjunct faculty members, were honored with ACM SIGCHI Special Recognitions. Peek was recognized for “democratizing automation through open-source hardware, building global maker communities and bridging academic research with grassroots fabrication practices,” and Aragon “for establishing human-centered data science as a new field bridging HCI and data science, demonstrating its impact through applications from astrophysics to energy systems.”

Read more about the 2025 ACM SIGCHI Awards, and see more about DUB and the UW presence at CHI 2025. 

Read more →

One of the foundations of database theory is efficient query execution. On the theory side, researchers have been tackling this issue by finding the upper bounds on the query size results to determine the fundamental hardness of query execution. Other researchers have been taking a practical approach by honing query optimizers that can automatically create query evaluation algorithms based on various data properties. These two approaches have converged in the form of worst-case optimal join algorithms, which ensure the optimal execution time for query evaluations by taking into account certain statistics about the queried dataset. 

In a paper titled “Partition Constraints for Conjunctive Queries: Bounds and Worst-Case Optimal Joins,” Allen School Ph.D. student Kyle Deeds introduced an innovative statistic called partition constraints that can improve existing worst-case optimal join algorithms by capturing the latent structure in relations within the data. The rigid nature of traditional constraints used in worst-case optimal join algorithms, such as degree constraints or cardinality constraints, can end up hindering query execution speed. Instead, partition constraints extend the notion of degree constraints, enabling the relationships between database attributes to be divided into smaller and more manageable sub-relations that each have their own degree constraints.

Deeds and his collaborator Timo Camillo Merkl of TU Wien presented their research at the 28th International Conference on Database Theory (ICDT) in Barcelona, Spain, last month where their work received the Best Student Paper and Best Paper Awards.

“A core problem of database theory is query execution,” said Deeds, who is advised by professors Dan Suciu and Magdalena Balazinska in the University of Washington’s Database Group. “What this work did was present a new kind of statistic, called partition constraints, on that data that we can take into account when we are executing a query, essentially speeding up the whole process for data that has this nice structure.”

An example of a use of partition constraints is in an algorithm that records who has key card access to which rooms within a university. Faculty and students may only need to enter a few rooms such as lecture halls and offices, while security and cleaning staff may need to be able to reach many more, or maybe all, rooms. It makes sense to then partition keycard clearance by tracking the access restrictions of faculty and students, and of the various room caretakers, to efficiently determine who needs access to which rooms.

Deeds took inspiration for partition constraints from another field of computer science called graph theory, or the study of networks. Degeneracy in graph theory measures how sparse a graph is. In a network with low degeneracy, researchers can design algorithms that minimize the performance impact of highly connected vertices, which can make some graph algorithms more efficient. When it comes to partitioning and databases, this property can also help maintain database performance, even in cases where the tables have very high degrees. The researchers developed partition constraints as a declarative version of graph degeneracy for higher-arity data, or data with more attributes or fields.

“This paper is like a translation style work,” Deeds said. “The goal of this work was to translate ideas from graph theory over in a way that made sense to the database community. Once we found the right way to do the translation, things just clicked into place.”

These ideas started clicking into place when Deeds and Merkl met and began collaborating on their research.

“Kyle and Camillo first met at a previous instance of the same conference, at ICDT in 2023 in Greece. Since then they worked remotely, without any help or guidance from their respective advisors,” said Suciu, who holds the Microsoft Endowed Professorship in the Allen School. “They started from an existing, elegant concept in graph theory, called ‘degeneracy,’ and extended it to a practical method for processing relational databases. Their main idea is that, by carefully partitioning each relation into a small number of fragments, then computing a query separately on each fragment, one can significantly reduce the query’s execution time.”

For Deeds, partition constraints point to a new direction for database theory research to pursue. Partition constraints help uncover underlying and useful structures within data, and understanding the different correlations and properties that a database has can help find ways to further optimize query executions.

“The same partitioning method that Kyle and Camillo developed has many other applications beyond query evaluation. For example, in one of my research projects we are developing improved techniques for cardinality estimation, and asked Kyle to join us to adapt his method for this problem,” Suciu said. 

Outside of partition constraints, Deeds has been researching ways to optimize other programs. He introduced Galley, a system for declarative sparse tensor computing that enables users to draft efficient tensor programs without needing to make complex algorithm decisions. 

Read the full paper on partition constraints. 

Read more →

Allen School professor Su-In Lee, who directs the University of Washington’s AI for bioMedical Sciences (AIMS) Lab, is shaping the future of biology and medicine through artificial intelligence. Her research focuses on fundamentally advancing AI and machine learning (ML) techniques to provide insights into complex biological systems and drive healthcare breakthroughs, transforming fields from basic biology to clinical medicine, including cancer biology, dermatology and critical care. 

Lee’s groundbreaking work has earned her a long list of accolades. Most recently, the International Society for Computational Biology (ISCB) inducted her into its 2025 Class of Fellows. These fellows lead the field with their innovative research and service, reflecting “a career of significant impact and a dedication to the scientific community.”

“I am really grateful and honored to be named an ISCB Fellow,” said Lee, who holds the Paul G. Allen Professorship in Computer Science & Engineering. “This recognition fuels my desire to contribute further and push the field to the next stage.”

One of Lee’s most pivotal contributions to the field is her work on the SHAP, or SHapley Additive exPlanations, values. The technique uses a game theory approach to help explain the output of an ML model. Her research using SHAP techniques tackles the accuracy versus interpretability problem. Simpler models can be more interpretable at the expense of being less accurate, but complex models are more accurate but difficult to interpret. Instead, Lee and her collaborators balance the two to develop high-performance, expressive models that are effective for biomedicine.

Using SHAP values as a foundation, Lee and her collaborators developed a novel framework called Prescience that could predict and also explain a patient’s risk of developing hypoxemia during surgery. Their follow-up research CoAI, also known as Cost-Aware Artificial Intelligence, used SHAP values to reduce the time, effort and resources needed to predict patient outcomes and inform the treatment plan. Lee has also used SHAP values to understand factors influencing aging to better treat age-related disorders using the ENABL Age, or ExplaiNAble BioLogical Age, framework. 

Lee has also developed explainable AI methods that can analyze biological data and find new therapeutic targets for diseases. She and her colleagues introduced ConstrastiveVI, a deep learning framework that applies contrastive analysis to single-cell datasets that help separate variations in the target cells from the healthy control cells during experiments. For Lee, explainable AI also has the potential to identify novel treatments for Alzheimer’s disease, which is one of the 10 leading causes of death in the U.S., and other neurodegenerative conditions. Lee has also used explainable AI to audit and dissect the reasoning process that AI systems use to analyze medical images, such as chest X-rays for predicting COVID-19 status or dermatological images for detecting melanoma.

“Su-In’s body of work has completely transformed how people analyze and interpret AI/ML models. Her SHAP approach is used by essentially the entire field, not only in computational biology but all of computer science,” said Trey Ideker, professor of medicine, bioengineering and computer science at the University of California San Diego.

In addition to her research, Lee has helped promote interdisciplinary education and collaboration. Since 2020, she has served as director of UW’s Computational Molecular Biology program. Lee has also played a pivotal role in establishing the Medical Scientist Training Program and Allen School collaboration to train Ph.D. students in both computer science and clinical medicine. She has also taken her leadership skills beyond UW as co-founder and co-chair of the Machine Learning in Computational Biology conference.

Lee’s fellowship follows on the heels of another ISCB recognition. For her contributions to AI and biomedicine, she received the prestigious 2024 ISCB Innovator Award, honoring leading mid-career computational biologists. Last year, she was the first woman to win the Samsung Ho-Am Prize Laureate in Engineering, often referred to as the “Korean Nobel Prize,” and was also named a Fellow of the American Institute for Medical and Biological Engineering. 

Having transitioned from studying computer science to biology herself, Lee understands the challenges Allen School students may face in researching biomedical applications, but also the rewards.

“It would be great for the field, the world and science in general if more computer science students pursue computational biology,” Lee said. 

Read more about this year’s ISCB Fellows and Lee’s research.

Read more →

When you look at a knitted panel head on, all you might see is gray noise, but taking a step to the side and looking at an angle reveals another image — a hidden Mona Lisa. 

This technique is called illusion knitting, where stitches of varying heights can create different images depending on what angle you view the piece. Traditionally, the intricate process of designing these illusions, by selecting stitches, choosing their colors and deciding whether they are raised or not was done by hand. Now, a team of Allen School researchers have introduced computational illusion knitting. The design framework helps automate the process by computationally generating knitting patterns for input designs, making illusion knitting more accessible and allowing for more complex and multi-view patterns that were previously impossible. The researchers presented the paper at the SIGGRAPH 2024 conference.

“If we write down the mathematical properties of the illusion knits, then we have a foundation with which to create algorithms that help the user iterate on and optimize their designs,” lead author and Allen School Ph.D. student Amy Zhu said. “With this foundation you can start to think, ‘where can I push the boundaries of what’s possible with illusion knitting?’”

To create an illusion knit object, the researchers first characterize the design’s unique microgeometry, or the small-scale geometry on the surface of a knitted piece. The goal is to then add a layer of abstraction over the complex microgeometry to streamline the design process, Zhu explained. This layer uses logical units called bumps, or raised sections made by a knit with a purl in the next row below it, that can block previous stitches, as well as flats, where the knit surface is level and is made by a knit stitch with another in the next row. 

These observations led the researchers to develop various constraints that capture both the viewing behavior, or what image is seen at each angle, and the physical behavior, also known as the result from the knitted design or fabrication choices. Single image illusion knits, such as the hidden Mona Lisa, are easier to design as they only change the color between rows.

Using computational techniques, the team became the first to design and execute illusion knitting patterns that require mixed colorwork and texture between rows. The researchers first used automated methods such as gradient descent and the maximum satisfiability problem (MaxSAT) to find a compromise that satisfies the most constraints and relaxes others, Zhu explained. However, these systems are still limited by how well they can express constraints about readability, knittability as well as how accurate the theoretical model is to reality.

Instead of relying solely on the algorithm for solutions, the researchers created a user-in-the-loop framework that allows the user to easily edit the design. The user can then further simplify the design by breaking it down into tiles made of bumps and flats to fill the image.

“There’s many tricky things to consider when you’re in fabrication land that are often difficult to anticipate or capture when working theoretically — maybe the machine settings are wrong and the white stitches are smaller, making the contrast between colors lower,” Zhu said. “One reason I like our approach is that we put the fabricated reality first, so our priority is providing ways for the user to edit things that come out of the machine, so they can observe if something didn’t look right and know how to fix it.”

For example, using a knitting machine and this design framework, Zhu created a knit piece that shows artist Vincent Van Gogh’s self portrait from one angle and his sunflowers from another. The quantized input image of the self portrait had too much noise around his head which made the design unreadable when knit. For a cleaner result, all she had to do was edit the input image to remove the noise or modify the tiles.

Zhu said she hopes this framework gives users the foundation to create more innovative illusion knit designs such as clothes that display images with certain poses or knit animations. Next, Zhu is researching ways to capture and explore different fabrication plans of knit objects.

“Approaching crafts from a computer science angle allows us to expand the boundaries of what can be achieved — whether that’s developing new algorithms for machine knitting, creating novel design tools or even inventing new forms of visual art,” said senior author and Allen School professor Zach Tatlock, who co-advises Zhu alongside colleague Adriana Schulz. “It’s fascinating to see how the rigor of computer science can enhance the creativity of traditional crafts.”

Additional authors include Allen School Ph.D. students in the Graphics and Imaging Laboratory (GRAIL) Yuxuan Mei and Benjamin Jones, along with Schulz.

Read the full paper on computational illusion knitting. 

Read more →