This summer, 10 University of Illinois PhD students spent 12 weeks working alongside IBM researchers through the IIDAI Summer Externship Program. Their projects spanned quantum computing, artificial intelligence, distributed systems, networking, IT automation, and advanced computing infrastructure.
Across the cohort, students gained firsthand experience with industrial research while contributing to ambitious technical projects. Their work included quantum-classical computing and quantum error correction, quantum simulation and chemistry, AI agents, GPU resource provisioning, networking systems, and new approaches to accelerate generative AI. Several projects resulted in papers, ongoing research collaborations, and plans for open-source development.
Just as important as the technical accomplishments was the opportunity to experience IBM's collaborative research environment. Students repeatedly highlighted the freedom to explore their own ideas, the accessibility and support of IBM researchers, and the opportunity to learn how research can move from an initial concept toward practical applications.
The experience also extended beyond individual research projects. Students built connections with researchers and fellow interns and externs through technical discussions, networking, social activities, sports, BBQs, and other gatherings. For several participants, those relationships and research collaborations are continuing beyond the 12-week program.
Together, the students' experiences demonstrate the value of the IIDAI partnership in connecting Illinois researchers with IBM expertise. The externship provides more than a summer research opportunity. It gives students new perspectives, professional networks, and research collaborations that can continue to grow long after they return to campus.
2026 Externship at a Glance
10 Illinois PhD Students | 12 Weeks | IBM Research | AI, Systems & Quantum Computing
· Research spanning quantum computing, AI, distributed systems, networking, IT automation, and computing infrastructure
· Projects resulting in papers, research outcomes, and continuing collaborations
· Hands-on experience working alongside IBM researchers and mentors
· Exposure to the process of translating research ideas into practical technologies and open-source tools
· New professional relationships connecting Illinois and IBM researchers beyond the summer
Participant Experiences
Ayush Bansal, PI: Prof. Tianyin Xu
"My time at IBM was an incredibly rewarding experience. It was a privilege to interact with so many brilliant minds in the Quantum division, and I am immensely grateful to my mentors, Apoorve Mohan and Shraddha Singh, for their guidance. The culture at IBM stands out; everyone is exceptionally humble, eager to share, learn, and always open to discussing new ideas.
Our collaboration was highly productive. We focused on quantum-classical pipelines and quantum error correction and recently submitted a paper to the SFW-SC26 workshop (SuperComputing 2026). Thanks to IIDAI, we are easily able to maintain this momentum and are already actively working together on the two projects."
Jiaqi Lou, PI: Nam Sung Kim.
"My externship at IBM was a great experience, both for the research and for the people I met. I especially enjoyed working with my mentors on an intra-host multi-path and topology project using real off-the-shelf hardware. It was exciting to work on a problem closely related to my PhD research, but from a different perspective and in an industrial research environment. I also had the chance to meet researchers from different teams, learn about their work, and hear about their career paths. I really enjoyed the collaborative research culture at IBM Research. Overall, the externship gave me new technical perspectives, valuable connections, and a better understanding of industrial research at IBM."
Jinghua Wang, PI: Deming Chen.
"In this summer, I worked on a multi-agent smart resource provisioning framework for modern Kubernetes GPU clusters, an IIDAI project that stretched me across both AI agent design and distributed systems. Outside of the research, one of my favorite moments was the IBM BBQ 2026 event: live music and singing, great BBQ, and plenty of delicious food to share. It was a wonderful opportunity to connect with my IBMer colleagues and make great memories together!"
Shubham Kumar, PI: Narendra Ahuja.
"The IIDAI externship shifted my perspective on research. My eyes were opened to the unique problems faced in industry, which are markedly different than what is typically pursued in academic settings. My mentors allowed me to shape my own research problem and helped guide me towards a promising solution.
Aside from research, it was a pleasure to interact with other PhD students, many from other institutions and research areas. Full-time folks at IBM were very energetic in organizing weekly sports events, and living with other interns and externs at PACE helped us grow close."
Kai-Siang Wang, PI: Indranil Gupta
"This summer, I had the opportunity to work at IBM on a project related to agentic workflows. During my time there, I learned a lot from the people around me and from other projects, especially about emerging topics in IT automation and how research ideas can be integrated into real-world products.
One of the things I enjoyed most about working at IBM was the relatively flat organizational structure. I could easily connect with managers and discuss my ideas and thoughts, which made the experience both engaging and rewarding. Beyond the work itself, the fountain drinks, BBQ events, and World Cup watch parties were also memorable parts of my summer at IBM!
The below picture is the friends I made during the summer!"
Simon Seymour, PI: Andre Schleife.
"Hey, my name is Simon, and I just finished the first year of my PhD in the Materials Science & Engineering Department at UIUC. Through the IIDAI partnership, I had the opportunity to work this summer in IBM’s Chicago office as part of the Quantum Algorithms Team, under the guidance of Kunal Sharma.
I really enjoyed my time there, both because Chicago in the summer is truly fantastic and because I was given the support and freedom to explore my own ideas and research interests. In particular, I mainly worked on improving semidefinite programming algorithms for solving the ground-state problem of quantum chemistry Hamiltonians. This class of algorithms also has much broader applicability, including within quantum error mitigation workflows.
There is such a talented and diverse team of researchers at IBM, and my supervisor helped me make the most of that environment by connecting me with people whose expertise could help me advance my research. By learning about the work of my peers, I was also exposed to new topics and ideas that will actively shape the remaining years of my PhD.
Overall, this experience has reinforced my ambition to develop and use computational methods, whether classical or quantum, to tackle real-world problems in materials science."
Siqi Yang, PI: Yuxiong Wang
"This summer, I had the opportunity to intern at IBM Research, where I explored an open-ended research question: when a coding agent is allowed to evolve its own harness—including its prompts, tools, and memory—what is it actually learning? One of the highlights of my internship was having the freedom to identify and pursue my own research idea. I worked with my mentors to shape the initial direction, design experiments, and investigate the problem across 8 programming languages and 3 base models. The results revealed surprisingly structured patterns in what self-evolving coding agents learn, and ultimately became our paper, “One Recipe, Many Harnesses: What Self-Evolution Encodes Across Languages and Models.”
Another especially rewarding part of the experience was being able to take the project from an initial idea all the way to a completed paper. Early in the internship, I set a personal goal of wrapping up the work before the end of the summer, and I was excited to see that goal become a reality. I am very grateful to my mentors, Saurabh Pujar and Martin Hirzel, for their guidance and thoughtful feedback, as well as to my co-authors for a great collaboration. Overall, my time at IBM Research gave me valuable experience in conducting independent research, turning an open-ended idea into a concrete project, and seeing it through to a final research outcome."
Yi-Ting Lee, PI: Andre Schleife
Throughout the 12 weeks of the summer, I had the opportunity to begin the final project before I defend my PhD. Together, we worked on the quantum simulation of Gibbs state dynamics and studied its potential applications. I also had the chance to collaborate with other researchers, which allowed me to learn new numerical techniques. Wrapping up my PhD with this project has been the best possible transition into my quantum career, and it's left me excited about where the field is heading.
Akash Vijay, PI: Jong Yeon Lee
My externship at IBM through the IIDAI Externship Program was an intellectually rewarding and inspiring experience. Working in Yorktown offered a sense of connection to the history of computing, alongside the opportunity to explore IBM’s cloud technologies and work with its quantum hardware. Meeting fellow interns and externs was another highlight, and I formed connections that I look forward to maintaining. My research with Nate Earnest-Noble focused on simulating quantum circuits with symmetries and realizing dynamical quantum error-correcting codes. Our collaboration is ongoing, and we hope to complete the project soon. The externship deepened my understanding of quantum computing and gave me valuable research experience that I look forward to building on in my future work.
Zheng Wang, PI: Minjia Zhang
This summer at IBM Research, I focused on understanding and addressing a key limitation of diffusion-based speculative decoding. We observed that diffusion-based draft models, due to their lack of explicit causal conditioning during parallel generation, can assign high probabilities to individually plausible tokens while still producing draft sequences that are not coherent under the autoregressive prefix. To address this issue, we developed XPress, which introduces a lightweight causal refiner to correct the logits produced by the diffusion drafter using causal context. We further leverage Jacobi Decoding to resolve the sequential dependencies introduced by causal refinement while maintaining efficient parallel generation. Our experiments show that XPress can consistently improve the acceptance length of existing diffusion drafters and translate these gains into higher end-to-end speculative decoding throughput. The project has also received strong internal support at IBM, and we are currently actively working toward integrating XPress into open-source training and inference frameworks to make the method more accessible and useful to the broader research community.
At IBM, I had the opportunity to work closely with researchers who were extremely supportive, open to discussion, and enthusiastic about exploring new ideas. Through frequent technical discussions and collaboration around the broader open-source integration effort, I learned a great deal about turning research ideas into practical and reusable tools. I especially appreciated the collaborative environment at IBM, where I was encouraged to explore challenging problems independently while always being able to receive valuable feedback and support from the team.