Workshop Program
June 22, 2026 — Dubois Center, UNC Charlotte Center City, Charlotte, NC, USA
| 8:30 – 9:00 |
Registration
|
| 9:00 – 10:30 |
Session 1: Opening & Red-Teaming Generative AI
|
|
Welcome to DSML 2026
Pietro Liguori and Zitao Chen, PC Chairs
|
|
Keynote: AI and Dependability in Computing Systems: Friend or Foe?
Saurabh Bagchi, Professor, Purdue University and KeyByte
Q&A
|
|
Trust Without Safeguards: A Red-Teaming Study of ProteinMPNN in Open Protein Design Pipelines
Tia Pope, Ahmad Patooghy (North Carolina A&T State University)
|
| 10:30 – 11:00 |
Coffee Break
|
| 11:00 – 12:30 |
Session 2: Trustworthy LLMs in Security-Critical Applications
|
|
Drivers of Secure and Correct Code: A Factorial Study of Size, Pre-Training, and Data Quality
Pietro Liguori (University of Naples Federico II), Rrezarta Krasniqi (University of North Carolina at Charlotte), Domenico Cotroneo (University of North Carolina at Charlotte)
|
|
LLM-PEA: Leveraging Large Language Models Against Phishing Email Attacks
Najmul Hasan (University of North Carolina at Pembroke), Prashanth Busireddygari (University of North Carolina at Pembroke), Haitao Zhao (University of North Carolina at Pembroke), Yihao Ren (University of North Carolina at Pembroke), Jinsheng Xu (North Carolina A&T State University), Shaohu Zhang (North Carolina A&T State University)
|
|
On the Reliability of Targeted Unlearning in 4-Bit Quantized LLMs [Online presentation]
Syed Ahsan Ali (NED University of Engineering and Technology)
|
|
Establishing Zero-Shot LLM Performance at Network Tomography
India-Jane Barry (City St George's, University of London), Ilir Gashi (City St George's, University of London), Kizito Salako (City St George's, University of London), Pedro Marques (BT), Pranava Madhyastha (City St George's, University of London)
|
|
Latent Space Probing for Adult Content Detection in Video Generative Models
Alizishaan Khatri (Wrynx Inc.), Chiquita Prabhu (Independent Researcher)
|
| 12:30 – 14:00 |
Lunch Break
|
| 14:00 – 15:30 |
Session 3: From Practice to Research — Dependable ML in Production
|
|
Keynote: When the Pager Goes Off: Dependable and Secure ML in Regulated Enterprise Operations
Pramod Muppala, Vice President and Middleware Team Lead, Enterprise Engineering & Technology, Bank of America
Q&A
|
|
AdROD: Adaptive Redundancy for Object Detection in Autonomous Driving
Shunsuke Nagao, Fumio Machida (University of Tsukuba)
|
|
[Research Proposal] Resilience Assessment of AI Accelerators: A Driver-Level Fault Injection Methodology
Marcello Cinque, Luigi De Simone, Nike Di Giacomo (University of Naples Federico II)
|
| 15:30 – 16:00 |
Coffee Break
|
| 16:00 – 17:30 |
Session 4: Federated Security & Model Robustness
|
|
On the Extreme Variance of Certified Local Robustness Across Model Seeds
Minh Le (Georgia Institute of Technology), Phuong Cao (University of Illinois Urbana-Champaign)
|
|
Model Chunking in Decentralized Learning: From Privacy Defense to Privacy Leakage
Halil Betmezoglu, Bart Cox, Jérémie Decouchant (Delft University of Technology)
|
|
Color Matters: Trigger Color Affects Success in Federated Backdoor Attack
Kavindu Herath, Joshua Zhao, Saurabh Bagchi (Purdue University)
|
|
Seeing in Shades of Gray: Real-time Federated Image-Based Malicious Traffic Detection
Sapthak Mohajon Turjya (North Carolina A&T State University), Moaz Usama (The British University in Egypt), Mulham Fawakherji (North Carolina A&T State University), Mahmoud Abouyoussef (North Carolina A&T State University), Islam Obaidat (North Carolina A&T State University)
|
|
FlowMamba: A Dependable and Predictable Streaming Graph Framework for Energy IoT Intrusion Detection
Chuang Huo, Gang Wang (Inner Mongolia University of Technology)
|
|
Closing Remarks
Pietro Liguori and Zitao Chen, PC Chairs
|
Saurabh Bagchi
Professor, Purdue University and KeyByte
AI and Dependability in Computing Systems: Friend or Foe?
Abstract: We look at two sides of the coin with respect to AI's effect on the reliability and security of computing systems. With respect to AI and reliability, in some aspects, it is a foe, for example, by making it more challenging to reason about probabilistic programs. On some other aspects, it is a friend, for example, by enabling causal analysis for root cause diagnosis of failures. With respect to AI and security, in some aspects, it is a foe, for example, by enabling phishing and vishing attacks at scale. On some other aspects, it is a friend, for example, by automatically finding and patching vulnerabilities in OSS.
He then talks of some recent developments in the NSF Center, which serve to highlight the role of AI as a friend in reliability and security. An interactive exercise will be conducted at the talk to understand the views of the audience.
Speaker Bio: Saurabh Bagchi is a Professor in the School of Electrical and Computer Engineering and the Department of Computer Science at Purdue University in West Lafayette, Indiana. His research interests are in dependable computing and distributed systems. He is the Director of the NSF CISE Center Chorus (2024-ongoing) and of the Army's Artificial Intelligence Innovation Institute (A2I2) (2020-ongoing). He was selected to the International Federation for Information Processing (IFIP) (2020) and is a Fellow of the International Academy of Artificial Intelligence Sciences (AAIS 2025) and of the Institute of Engineering and Technology (IET 2022). Saurabh is the Founder of a cloud computing startup, KeyByte.
Saurabh is proudest of the 27 PhD students and 50 Master's thesis students who have graduated from his research group and who are in various stages of building wonderful careers in industry or academia. In his group, he and his students have way too much fun building and breaking real systems. Along the way, this has led to 13 best paper awards or runners-up awards at IEEE/ACM conferences and two Test of Time Awards. Saurabh received his MS and PhD degrees from the University of Illinois at Urbana-Champaign and his BS degree from the Indian Institute of Technology Kharagpur, all in Computer Science.
|
Pramod Muppala
Vice President and Middleware Team Lead, Enterprise Engineering & Technology, Bank of America
When the Pager Goes Off: Dependable and Secure ML in Regulated Enterprise Operations
Abstract: Much of the published work on dependable and secure ML assumes a degree of control over the data, the model, and the deployment environment that simply does not exist inside a large, regulated financial institution. In production, models share infrastructure with decades-old middleware, inherit the failure modes of the systems that feed them, and operate under audit, regulatory, and uptime constraints that academic threat models rarely capture.
This keynote offers a practitioner's view from inside one such environment. It walks through the operational realities of running ML alongside critical enterprise middleware: how dependability is actually measured and contested, where security and resilience requirements collide with model-development velocity, and which failure modes turn out to matter once a system is live. It draws on concrete examples from infrastructure automation, secure certificate and backup management, and the enablement of AI/ML workloads at scale, and ends with a set of open problems that the speaker believes the research community is well positioned to address but largely is not yet working on.
Speaker Bio: Pramod Muppala is Vice President and Middleware Team Lead within Enterprise Engineering & Technology at Bank of America, where he leads infrastructure automation, operational resilience, and secure AI/ML enablement across large-scale regulated production environments. He is a Senior Member of the IEEE and a co-inventor on multiple patent filings in the areas of secure system backups and automated SSL/TLS certificate discovery and management. His work focuses on the intersection of dependability, security, and ML in enterprise systems, with particular emphasis on what survives the transition from research prototype to regulated production.
|
|
|
