This summer marked the fifth year of the OpenDP Interns and Fellows Program! We received a record number of applications this year, and in June we welcomed three interns and two fellows.
Our internship program offers a summer of research and mentorship for students interested in implementing differential privacy. Guided by OpenDP’s vision, we match PhD and undergraduate interns with projects working side by side with our team. This summer our interns explored linear regression, differentially private principal components analysis, and usability for DP Wizard.
Each year, we also invite established professionals from industry and academia to join us as visiting fellows. Our fellows bring their existing projects to OpenDP and we pair them with hosts to support their work. This year, fellows joined us from Deloitte and the Swiss Federal Statistical Office, collaborating with us on projects related to federated health data and measurement error.
Read on to hear more from our 2025 interns and fellows!
Tell us a little about yourself:
I am a computer science PhD student co-advised by Adam Smith and Sofya Raskhodnikova at Boston University. My research focuses on differential privacy and sublinear time algorithms. Outside of research, I spend a lot of my free time reading, particularly fiction and sci-fi, but I also enjoy playing backgammon and biking.
What are you currently working on?
I am working on methods for private high-dimensional linear regression that can be implemented in practice. Existing approaches are either computationally intractable, incur suboptimal dependence on the dimension of the data, or achieve optimal dependence on the dimension but have significant overhead making them infeasible in practice. The goal of the project is to study realistic settings where the aforementioned issues can be circumvented.
How is that impactful for OpenDP?
Currently the OpenDP library only has a one-dimensional linear regression algorithm. The ultimate goal of the project is to design an algorithm for high-dimensional linear regression that can be integrated into the library.
What are you hoping to do after graduation?
I’m not sure exactly what I’d like to do yet, but I at least hope to keep solving interesting problems.
Onyinye Dibia
Tell us a little about yourself:
I’m a PhD student in Computer Science at the University of Vermont, originally from Issele-Uku, a small town in southern Nigeria. My background and experiences have shaped my passion for privacy tools that are not only technically sound but also usable. My research sits at the intersection of privacy, usability, and human-centered design. I focus on making differential privacy (DP) more understandable and accessible to a broad range of users, especially non-experts. This summer, I’m an intern with the OpenDP project, where I’m continuing that work. Outside of work, I enjoy cooking and experimenting with new dishes, mentoring, and exploring new places—Boston is a fun summer change of pace!
What are you currently working on?
This summer, I’m evaluating the usability of the DP Wizard, a user interface (UI) for building differentially private analyses using the OpenDP library. I’m conducting a user study with data practitioners to understand whether the Wizard helps them bridge the gap between a user interface and the OpenDP API. I’m also identifying where users struggle, especially during the transition from UI to code, and improving documentation and guidance to reduce friction for new users.
How is that impactful for OpenDP?
By focusing on real user experiences, my work aims to lower the barrier to entry for using the OpenDP library, especially for those without deep privacy expertise. Usability insights from the study will inform future improvements to the DP Wizard and the OpenDP documentation, helping OpenDP better support a broader community of analysts, researchers, and developers who want to adopt differential privacy in practice.
What are you hoping to do after graduation?
After graduation, I hope to continue research in usable privacy—whether in academia or a research-oriented industry role—focusing on making privacy technologies more intuitive, accessible, and impactful. My goal is to help build systems that not only offer strong privacy guarantees but also support a wide range of users, from data analysts to policymakers, in understanding and applying them effectively.
Rita Ionides
Tell us a little about yourself:
I come from the University of Michigan’s Department of Statistics, where my advisor is Ambuj Tewari. My research interests are in high-dimensional inference, statistical machine learning, and of course differential privacy. I became interested in differential privacy because of my background in biotech— I used to work in statistical genomics and pharmacology at University of Cambridge and University of Michigan Medical School. While I was there, I became curious about how methodology that is theoretically “privacy-preserving” can be not very private at all: a few years later, here I am at OpenDP!
Outside of research, I love dancing (salsa, bachata, ballet), visiting national parks, and absolutely demolishing the competition at trivia night. I also host a blog, Regular Expressions, where I write about differential privacy, CS theory, and generally being a student.
What are you currently working on?
I’ve been working on a new algorithm for performing differentially private principal components analysis (PCA).
How is that impactful for OpenDP?
PCA is a foundational tool for dimensionality reduction. It’s widely used in genomics, image compression, and feature selection, with critical applications in (among many others) healthcare, finance, and computer vision. Clearly, it’s an important tool for OpenDP’s users; and, as it turns out, the current version in the library has been leaking information through floating-point error. Correcting that gap— and then rigorously proving that the gap is closed— has been my focus for the last several months.
What are you hoping to do after graduation?
I’m still figuring out where exactly I’ll end up, and would welcome advice!
Lancelot Marti
Tell us a little about yourself:
I am currently working as a Data Scientist at the Swiss Federal Statistical Office. Before that, I was a Data Engineer at a startup in Geneva, where we developed an application to support real estate professionals. I hold a Bachelor’s degree in Economics and a Master’s in Business Analytics, which is where I developed my passion for data analysis. In my free time, I love exploring the Swiss mountains, whether it’s hiking or doing via ferrata.
What are you currently working on?
In my current role at the Swiss Federal Statistical Office, I support various Swiss public administrations with their data science projects. Within my team, we’re also developing a platform called Lomas, which will enable researchers to perform data science on sensitive data while ensuring privacy. To protect individuals’ information, we rely on libraries using differential privacy, one of them being OpenDP.
What drew you to become an OpenDP Fellow?
I was interested by the fact that our Lomas platform already uses the OpenDP library, so becoming an OpenDP Fellow felt like a natural way to deepen my understanding and contribute back to the community. I am especially interested in how the noise added for differential privacy can bias results if not properly accounted for. I would like to explore how we can adapt existing methods from the measurement error literature to address this challenge and apply them in the context of differential privacy.
What are you hoping to achieve with OpenDP this summer?
This summer, I hope to evaluate how Differential Privacy noise affects real-world analyses by comparing true and noisy datasets. My goal is to develop clear, practical notebooks that show the community what methods or tools they can use to handle the measurement error introduced by DP. If time allows, I’d also like to add insights from an additional case study with a Swiss university, but my priority is to create reusable examples that help others understand and manage DP’s impact on statistical results.
Mohammad Manzari
Tell us a little about yourself:
I am an R&D program manager at Deloitte Consulting, where I focus on building secure and privacy-preserving AI systems. I’m passionate about applications of statistics and machine learning in healthcare, especially enabling collaborative research and clinical decision-making.
What are you currently working on?
I’m currently working on researching applications of differential privacy for federated computing systems. More specifically, I am exploring the potential benefits and challenges of implementing differentially private cohort discovery mechanisms in federated health data networks.
What drew you to become an OpenDP Fellow?
I was interested in joining the OpenDP fellowship program because of the opportunity to meet and learn from some of the world’s leading experts in data privacy. In addition, the fellowship program seemed like a great opportunity to explore and share ideas in a collegial and intellectually open environment.
What are you hoping to achieve with OpenDP this summer?
I’m hoping to deepen my understanding about some of the core concepts underpinning differential privacy, as well as its limitations and how it can be used to advance collaborative research using sensitive medical data.
Interested in joining us next year? Join our Slack and mailing list so you’ll be the first to know when we announce the 2026 summer program. Thank you to all our 2025 interns, fellows, and mentors for a great summer!