Brown Systems Week
June 15–20, 2026 | Brown University, Providence, RI

Overview

The Systems Week is a five-day seminar on the principles and practice of software systems. Its goal is to prepare a cohort of 25–30 participants for cutting-edge research in the area of systems. Yearly topics combine recurring systems themes, combined with each year's special theme driven by broader trends in the field. Each topic is co-taught by a faculty pair with combined expertise on the subject, offering a chance to study a problem from complementary perspectives. A key goal of the seminar is to foster collaborations between faculty and students that far exceed the seminar's timeline, ideally into the rest of the summer and beyond.

Apply now for 2026!
Application deadline: April 1, 2026

Organizers

Keynote: An Exploration of Systems Security for Agentic AI

Lily Tsai, Google Systems Research  Group (SRG)

Abstract. AI agents, the newest class of systems embraced by the tech industry, introduce incredible security and privacy risks. Without the proper defenses, an adversary can exfiltrate or corrupt any data or system state available to an agent system using attacks like direct or indirect prompt injection, injecting natural language instructions into the agent model's context that cause it to perform adversary-controlled actions. Today, performing this new form of "remote code execution" attack on an AI agent requires little effort on the part of the adversary, making better system defenses essential.

This talk begins with background on AI agents, particularly focusing on how AI agent security differs from that of non-AI entities, human or traditional software systems. We will then cover an overview of when, and why, we can apply existing system security approaches, e.g., sandboxing, and where we need novel solutions and what those solutions might be, e.g., contextual yet deterministic defenses.

Lily Tsai

Bio. Lily Tsai works as a researcher and engineer at SystemsResearch@Google (SRG), currently investigating frameworks for better security and privacy in agentic systems. In 2024, Lily earned her PhD from MIT, advised by Frans Kaashoek in the PDOS group and Malte Schwarzkopf in the Brown ETOS group, where her research focused on systems for better data protection and security in web applications. Beyond data privacy and security, Lily is also broadly interested in multicore performance and scalability, and the application of formal methods in systems. Besides research, Lily loves playing violin, reading, hiking, climbing, and exploring the world around her!

Topics

Lectures and corresponding lab sessions fall under (1) a common systems core, and (2) each year's special theme, driven by broader trends in the field. Lectures are provided by a combination of systems faculty at BrownCS and guest faculty from other institutions. The Systems Week also includes one or more keynotes, faculty panels on relevant topics, ample opportunities for interaction with other participants, and numerous social activities.

A typical Systems Week will include five mini-courses, delivered in daily lectures from multiple faculty members. Example topics include automated parallelization and distribution, retrofitting privacy guarantees, automated reasoning for systems, and scalable dataflow and streaming systems. The target audience includes advanced undergraduate or early PhD students with a serious (and broad) background in systems, but not necessarily on any of the seminar's theme topics. An anticipated set of prerequisites and prior knowledge will be made available several weeks prior to the seminar.

Organization & Schedule

The sessions are non-overlapping, so all participants will have the opportunity to attend all lectures. Each lecture is 70–80 minutes, including Q&A, and each topic comprises a total of 4 lectures spread throughout the week. Most lectures will be given between 9am and 5:50pm during weekdays, during the following windows: Session A (9–10:20am), B (10:30–11:50am), C (1:30–2:50pm), D (3:00–4:20pm), and E (4:30–5:50pm). Courses will include hacking sessions, with ample opportunities for interactivity, and in-class assignments optionally to be continued at home.

Background sessions might be offered the weekend preceding the Systems Week; and, depending on participant and faculty interest, targeted research opportunities may be explored the weekend following the Systems Week.

The following is a sample schedule. Lectures and hacking sessions will take place in Brown's main campus, in Providence, RI. Example locations include the Watson Center for Information Technology (CIT), the Brown Data Science Institute (DSI), or the Institute for Computational and Experimental Research in Mathematics (ICERM).

The following schedule is a sample and subject to change.

Monday Tuesday Wednesday Thursday Friday
A: 9–10:20am ReasonRange GPUSys SupChain Analysis Analysis
B: 10:30–11:50am GPUSys ReasonRange Analysis GPUSys eBPF
Lunch Lily Tsai Keynote 🍱 🍕 Panel1 🍜 🌮 Panel2 🥗
C: 1:30–2:50pm RustPriv SupChain eBPF RustPriv GPUSys
D: 3:00–4:20pm Analysis eBPF RustPriv ReasonRange SupChain
E: 4:30–5:50pm eBPF RustPriv ReasonRange SupChain —
Dinner 🍛 🍕 🍜 🌮 🍣


Below is a preliminary list of minicourses for 2026:

  • Extensible OSes, Network Security, & eBPF (eBPF): A mini-course from Akshay Narayan and Nick DeMarinis on verifiable kernel extensions with eBPF.
  • Modeling GPU Performance (GPUSys): A mini-course on modelling and harnessing the power of GPUs for modern workloads by Milda Zizyte and Deepti Raghavan.
  • Holistic supply-chain security (SupChain): A mini-course on securing against third-party components from Nikos Vasilakis and Vasileios Kemerlis.
  • A range of reasoning for systems (ReasonRange): A mini-course from Tim Nelson and Rob Lewis on applying a range of manual-automated reasoning to systems.
  • Analyzing Rust to Guarantee Privacy (RustPriv): A minicourse from Malte Schwarzkopf and Will Crichton on privacy analyses for Rust programs.
  • Analyzing Shell Programs (Analysis): A mini-course on static and dynamic analysis and transformation of Shell programs by Nikos Vasilakis and Lukas Lazarek.

The Systems Week will also feature a keynote and a panel discussion on Grad School applications.

The Systems Week's two-faculty-per-minicourse setup is intended to enable fruitful interactions between faculty that would benefit from such interactions and who might not have such interactions already in place. With 320 minutes of class time and daily research-oriented assignments, each minicourse should provide focused time, students, and resources to identify a common research problem (or set of problems) in which to make a serious dent. The intention is for these interactions to go beyond the boundaries of the seminar.

Costs

Attendees have all of their living expenses covered—e.g., accommodation, food, and transportation to various activities—and should be able to engage meaningfully in research soon after the start of the seminar. Attendees are expected to attend all lectures during the day, with limited homework (1–2h) in the evening. Outside lectures and homework, the seminar combines research-oriented activities (e.g., a "Research Highlights" reception) with social activities (e.g., art nights, hikes around RI) and other organized events.

Apply!

The application process includes:

  1. A letter of motivation, where applicants outline their technical background, their anticipated gains from participating in the Systems Week, and optionally what they hope to contribute.
  2. Recommender name and contact information for people who can comment on the candidate's design, building, or characterization abilities in software systems, broadly construed.

Recommendation letters are only required upon request. Most applicants will not be asked to submit one, but please provide recommender name in the form in case a letter is needed.

An anticipated set of prerequisites and prior knowledge will be made available several weeks prior to the seminar.

Please use this form to submit your application materials. The deadline for applying is April 1st. If you have any questions about the form or Brown Systems Week 2026, email Nikos Vasilakis with subject line BSW'26.

Participants are expected to bring their own laptops. As software requirements vary between minicourses, the combined requirements will be announced several weeks in advance—along with a Discord server and mailing list for additional support.

Each participant will be assigned a point of contact out of several volunteers helping with the organization of the Systems Week. Collectively, these volunteers—typically Ph.D. students at Brown CS—will also organize social events with research groups already at Brown, fostering deeper collaborations and friendships.

Brown Systems Week 2026 is made possible with support from Brown University, Brown Systems, the Brown Data Science Institute, the National Science Foundation, Google, and Amazon.

Brown University logo
Brown
Brown Systems logo
Brown Systems
Brown Data Science Institute logo
Brown Data Science Institute
National Science Foundation logo
NSF
Google logo
Google
Amazon logo
Amazon