Saranya Vijayakumar
Ph.D. Candidate in Computer Science
Carnegie Mellon University
saranyav [at] andrew.cmu.edu
Research Vision
I study the safety and security of AI systems under adversarial conditions, from single-model robustness to multi-agent coordination. My thesis develops structured, verifiable ways for increasingly autonomous AI systems to check claims and enforce consistency, spanning red-teaming, prover-verifier oversight, and neurosymbolic reasoning.
I am a PhD candidate in Computer Science at Carnegie Mellon University, advised by Christos Faloutsos and Matt Fredrikson, and an Anthropic Security Fellow. Before my PhD, I earned my A.B. at Harvard, writing my thesis with Cynthia Dwork and Jim Waldo, and spent three years as an associate at Goldman Sachs. My research spans formal methods, LLM alignment, and large-scale fraud detection, with collaborations at Inria, IBM Research, and Fujitsu. My work has been honored with Best Paper Awards, a NeurIPS Spotlight, and the 2026 ML and Systems Rising Star Award, and supported by the DoD NDSEG Fellowship.
Research
Sequential Pattern Recognition Attacks against Deployed Topic-Based Mechanisms
ICISSP 2026 (Best Student Paper Award)Saranya Vijayakumar, Matt Fredrikson, Norman Sadeh
Prototype-Integrated Representation Learning for Novelty Detection
IEEE TrustCom 2025Saranya Vijayakumar, Matt Fredrikson, Christos Faloutsos
PDFAICodeDetect: A Pipeline for Systematic Detection and Analysis of AI-Generated Code
IEEE MCSI 2025Saranya Vijayakumar, Philip Negrin, Christos Faloutsos
PDFMechanistically Interpreting a Transformer-based 2-SAT Solver
ICML 2025Nils Palumbo, Ravi Mangal, Zifan Wang, Saranya Vijayakumar, Corina Pasareanau, Somesh Jha
PDFAligned LLMs Are Not Aligned Browser Agents
ICLR 2025Priyanshu Kumar, Saranya Vijayakumar, Elaine Lau, Tu Trinh, Zifan Wang, Matt Fredrikson
PDF
Talks
Sequential Pattern Recognition Attacks Against Deployed Topic-Based Mechanisms
Current Topics in Privacy Seminar, March 31, 2026
- Security vulnerabilities in topic-based mechanisms
- Sequential pattern recognition attack methodology
LLM Security Vulnerabilities
AI Governance Course (17-416/17-716), March 31, 2025
- Applied security analysis techniques
- Real-world privacy challenges with LLMs
- Jailbreaking and watermarking
Security and Privacy in Practice
Information Security, Privacy & Policy (17-331/631), November 21, 2024
Slides- Applied security analysis techniques
- Real-world privacy challenges with LLMs
- Jailbreaking and watermarking
AI Security and Governance
AI Governance Course (17-416/17-716), April 3, 2024
- Current landscape of AI security challenges
- Intersection of technical capabilities and governance frameworks
- Emerging threats and mitigation strategies
AI Security, Robustness, and Privacy
Information Security, Privacy & Policy (17-331/631), December 5, 2023
- Overview of current challenges in AI security
- Discussion of robustness techniques and evaluation
- Privacy considerations in modern AI systems
Academic Service
Information Security, Privacy & Policy (17-331/631), Fall 2024
Final Project Judge
- Evaluated student projects on security and privacy implementations
- Provided technical feedback and industry-relevant insights
- Helped assess practical applicability of security solutions
Teaching
Teaching Philosophy
Teaching Statement PDFI believe in creating an inclusive learning environment that emphasizes practical understanding and critical thinking. My teaching approach combines theoretical foundations with hands-on experience, preparing students for both academic and industry challenges.
Teaching Experience
TA, Information Security, Privacy & Policy (17-331/631) — with Norman Sadeh & Hana Habib
Fall 2023TA, Rapid Prototyping (15-294 & 15-394) — with Dave Touretzky
Spring 2023Eberly Center Future Faculty Program — teaching development
2021 - 2023Mentorship
- Mentored high school student Philip Negrin on AI code-detection research (project video)
- Guided a team of 4 CMU Masters students on privacy research analyzing Google's Topics API (USENIX PEPR '24)
Research Projects & Impact
A few themes that cut across my work. Full details are in my CV and publications.
Agentic AI Security & Alignment
Red-teaming and alignment of LLMs deployed as autonomous agents. Showed that refusal-trained LLMs are easily jailbroken as browser agents (ICLR 2025), and now work on prover-verifier oversight and interpretability of agentic systems.
Neurosymbolic & Formal Methods
Grounding neural inference with SMT solvers (NeurIPS 2023 Spotlight), mechanistically interpreting transformer-based SAT solvers (ICML 2025), and formal verification of a secure messaging protocol used by the French government (Inria).
Large-Scale Fraud Detection
Scalable visualization and anomaly-detection methods for million-scale telecom and credit-card graphs, yielding deployment-ready systems with Mobileum (CIKM 2023, IEEE Big Data 2022, AAAI-23 demo).
Public-Interest & Policy Research
Earlier work spanning algorithmic fairness (Harvard Political Review, undergraduate thesis), a worldwide encryption survey with Bruce Schneier (covered by NBC News, The Intercept), and a CSET review on cybersecurity risks of AI-generated code for Georgetown.
Service & Leadership
Academic Community Building
Outreach & Impact
Implemented healthcare technology solutions with Partners in Health, Lima, Peru
Led computer science education programs in Boston public schools
Selected Media Coverage
Taking on Tuberculosis — Harvard SEAS
Trying to Stop Encryption? It's Everywhere — NBC News
New Survey Suggests U.S. Encryption Ban Would Just Send Market Overseas — The Intercept
Bringing Computer Skills to Classrooms — Harvard Gazette
Awards & Recognition
ML and Systems Rising Star Award
Anthropic Security Fellow
Graduate Fellowship for STEM Diversity (GFSD)
NSA (Declined)
National Defense Science & Engineering Graduate Fellowship
Army Research Office
Future Faculty Program
Carnegie Mellon University Eberly Center
Research & Industry Experience
Research Intern, Agentic AI
Jan - May 2026Fujitsu Research - Pittsburgh, PA
- Studying red-teaming of agentic AI with Koichi Onoue and Roshni Kaushik
- Architecting, building, and deploying LLM-powered and agent-based systems end-to-end
Research Intern, Trustworthy AI
May - Aug 2025IBM Research - Yorktown Heights, NY
- Developed a game-theoretic extension of prover-verifier games for legibility using multiple specialized agentic verifiers
- Advised by Erik Miehling and Karthikeyan Ramamurthy; emphasis on foundational LLMs, agentic AI, and reinforcement learning (RLHF)
Visiting Scholar, PESTO (Security Protocols)
Oct - Nov 2024Inria - Nancy, France
- Formal verification of the security properties and transcript consistency of a secure messaging platform used by the French government
- Advised by Charlie Jacomme and Steve Kremer
Research Collaborator, AIDA Platform
2021 - 2026Mobileum - Braga, Portugal
- Collaborated with Mobileum, a global telecom analytics provider serving over 900 operators worldwide, on industry-scale fraud detection research
- Analyzed real-world call graph data to develop scalable anomaly detection techniques
- Published multiple peer-reviewed papers and demos, including deployment-ready systems for detecting telecom fraud on million-scale graphs
Earlier Experience
Data Scientist, Electronic Trading (GSET) — Goldman Sachs
2018 - 2021Data Scientist, Distributed Organizing — Beto O'Rourke for U.S. Senate
Summer 2018Cybersecurity Intern — Booz Allen Hamilton
Summer 2017