Copi-Role | Yao Zheng@UHM

DOE Genesis Mission: STRATOS: Security and Trust Runtime Architecture for Time-critical Operational Science

Mon, 17 Aug 2026 10:10:00 -1000

Executive Summary

STRATOS develops a cloud-native, model-agnostic security middleware for protecting AI models used in time-critical scientific and energy-system operations. AI is increasingly embedded in workflows such as grid forecasting, contingency analysis, and operational decision making, creating new risks from adversarial perturbations, inference-time evasion, and backdoor attacks that may remain physically plausible while corrupting model outputs. STRATOS addresses this problem through a physics-grounded DevSecOps architecture that continuously evaluates inference streams, computes certified trust scores, and distinguishes malicious manipulation from legitimate operational variability without requiring modification of the protected model.

The project combines physics-constrained adversarial emulation, imbalance-resilient certified detection, and a multi-tier runtime mitigation pipeline that includes targeted input purification, Control Barrier Function-based graceful degradation, and quarantine or rollback of compromised data and models. The Phase I system will be evaluated across the University of Hawaiʻi at Mānoa campus microgrid and Argonne National Laboratory’s Controller-Hardware-in-the-Loop platform, targeting at least 90% certified detection, no more than 5% false positives, at least 95% interception of synthesized adversarial payloads, and end-to-end latency of 20 ms or less. The longer-term goal is to transition STRATOS toward a federated security architecture for trustworthy AI across DOE scientific computing and critical-infrastructure environments.

CyberAI Innovation: Securing Artificial Intelligence Agents: A Scenario-Based Educational Platform for Future Cybersecurity Professionals

Sat, 15 Aug 2026 15:29:00 -1000

Executive Summary

AI agents that autonomously browse the web, manage email, execute code, and query databases are being deployed across federal agencies at accelerating pace. These agents create a qualitatively new attack surface: a poisoned email can cause an agent to exfiltrate files; a malicious code comment can redirect an execution pipeline; a hidden webpage instruction can hijack a browsing session. Yet cybersecurity education has not kept pace. Existing AI security tools (Lakera Gandalf, OWASP LLM labs, CTF competitions) teach only chatbot-level prompt injection, a single-turn, text-only model that does not prepare students for agent-mediated threats involving tool access, autonomous action, and multi-step reasoning chains.

This project develops AgentSec, a scenario-based educational platform that teaches undergraduate cybersecurity students to identify, exploit, and defend against threats unique to autonomous AI agents, organized into three thrusts. Thrust 1 creates the educational framework, where a trust boundary taxonomy organizes agent vulnerabilities into three progressive categories, i.e., data ingestion, tool-action, and reasoning chain boundaries. This gives students a transferable mental model. Based on the taxonomy, we design a three-module curriculum with six organization-contextualized scenarios, red-team/blue-team experiential learning cycles, and a configurable difficulty framework. Thrust 2 builds the educational technology, where we develop sandboxed environments to enable students’ interaction with real LLM-powered agents with controlled tool access, a layered assessment system with an AI tutoring agent, an instructor analytics dashboard, and a scenario authoring toolkit for community-contributed content. Thrust 3 evaluates impact through a quasi-experimental study, assessing platform quality, student learning, and workforce placement into government and industry CyberAI roles. All code will be open-source, self-hostable via Docker Compose with open-weight models, and released through CLARK for nationwide adoption.

NVIDIA Academic Grant Program: OmniPort: Real-Time Semantic Digital Twins via ISAC and AI-RAN for Smart Ports

Sat, 15 Aug 2026 15:29:00 -1000

Executive Summary

OmniPort develops a real-time intelligent wireless and digital-twin platform for safer and more efficient autonomous operations in container ports. Ports are particularly challenging environments for robots because stacked metal containers obstruct GPS and line-of-sight sensors while producing strong wireless multipath. Instead of treating these reflections solely as interference, OmniPort uses integrated sensing and communication (ISAC) to extract information from routine robot uplink signals, enabling around-corner hazard detection and GPS-free robot localization. An AI-RAN then combines these sensing results with robot mobility and network telemetry to predict wireless conditions, dynamically allocate network resources, and maintain low-latency communication for safety-critical operations.

The project integrates these capabilities into an uncertainty-aware semantic digital twin that represents container geometry, robot movement, hazards, and wireless connectivity in real time. Rather than continuously transmitting camera video, operators can supervise robot fleets through a privacy-preserving VR environment generated from this semantic information, with warnings for collision risks, connectivity degradation, and emerging hazards. The platform uses four on-premises NVIDIA RTX PRO 6000 GPUs to support concurrent ISAC processing, AI-RAN control, NVIDIA Isaac Sim/Cosmos-based digital-twin reasoning, and Omniverse VR visualization, targeting an end-to-end control latency below 50 ms. The resulting technologies, software, and datasets are intended to provide a foundation for intelligent robotic operations in Hawaiʻi’s ports and other complex industrial environments.

Wearable Sweat Sensors for Cattle Heat Resistance and Metabolomics

Fri, 07 Nov 2025 00:00:00 -1000

Executive Summary

This CTAHR CARES pilot project develops non-invasive wearable sweat sensors to monitor heat stress in beef cattle at the molecular level. By coupling commercial sweat and internal temperature sensors with sweat metabolomics and gut microbiome profiling, the team aims to uncover biomarkers that distinguish heat-stressed from heat-resistant animals.

Heat stress in cattle is typically assessed using a combination of visual observation, physiological measurements, and environmental indices such as the Temperature–Humidity Index (THI). These methods are indirect, labor intensive, and often fail to capture individual animal differences in tolerance. This project will generate continuous, individual-level data on sweating, body temperature, movement, and molecular signatures to enable climate-resilient herd management and breeding strategies in Hawaiʻi and beyond.

Problem Statement

Beef cattle in tropical and subtropical environments face increasingly frequent heat-stress events driven by climate change. Current evaluation methods:

Rely on THI and farm-level weather data
Use intermittent measurements of body temperature and respiration
Depend on subjective observation of panting, drooling, and behavior

However:

Individual and breed variation in heat tolerance is large
Early, sub-clinical heat stress is often missed
Invasive or sporadic measurements are not scalable for large herds

There is a critical need for fast, accurate, and non-invasive tools to determine when a specific cow is experiencing heat stress, and to identify animals that remain resilient under extreme conditions.

Research Objectives

Overall Goal: Develop an integrated framework that links wearable sweat and temperature sensing with metabolomics and microbiome data to quantify heat stress and heat resistance in cattle.

Objective 1 – Characterize Heat Stress Physiology

Relate sweating rate, internal body temperature, and movement patterns to environmental conditions (THI, day/night cycles, seasonal variation)
Capture how apocrine sweat glands in cattle respond under different levels of heat load

Objective 2 – Identify Molecular Biomarkers

Perform sweat metabolomics to identify compounds associated with heat stress vs. heat resistance
Analyze gut microbiota (from fecal samples) and blood-based markers
Measure thyroid hormone levels, heat shock proteins (HSPs), and mitochondrial ATP-related gene expression as potential indicators of chronic heat load

Objective 3 – Enable Climate-Resilient Cattle Selection

Integrate physiological and molecular data to identify candidate biomarkers and genetic markers of heat stress and heat resistance
Provide a foundation for future selection and breeding programs targeting climate-resilient cattle in Hawaiʻi’s diverse environments