Anomalous activity detection for domain-specific heterogeneous SoCs

• NoC-based hardware counters and semi-supervised machine learning models for detecting anomalous activity resulting in compromised system availability
• DoS attack detection in representative implementation of SoC for communication subsystem in connected autonomous vehicles

Localizing detected anomalous activity in domain-specific heterogeneous SoCs

• Explainable machine learning techniques to extract impacted SoC locations based on hardware counter data used for detection

• EPOCHS: agile methodology for designing and implementing domain-specific SoCs for real-time cognitive decision engines in connected autonomous vehicles
• Anomalous activity detection: collected hardware performance data for analysis with machine learning techniques to detect compromised system availability
• Developed microbenchmark tests for reporting accelerator performance on EPOCHS FPGA prototype

• CHERI: replaces integer pointers with capabilities equipped with permission and bounds information, inherently offering spatial memory safety
• Memory versioning for temporal safety: reduces frequency of full memory sweeps required for capability revocation when freeing memory by 2n×, where n is the number of version bits
• Formal security analysis: Proposed a formal hardware-software contract for secure speculation on CHERI + memory versioning systems and presented a mathematical model of the system

• Apply CheckMate tool to an Intel Sandy Bridge-like microarchitectural specification written in Alloy to synthesize TLB exploits
• Synthesize candidate execution programs that account for hardware and OS operations arising from address translations
• Develop a framework to formally define a transistency model for x86 processors using hardware testing and verification techniques

• Vein imaging demo software: developed MFC application with capability to display image(s) from board level scientific camera, view histogram data of 12-bit grayscale pixel values, adjust contrast of image, and print image
• Wireless oscilloscope: high level design for wireless oscilloscope – analog input from Multi-Pixel Photon Counter (MPPC) module to digital output on Android application

• Investigated swarming strategies and potential tradeoffs for autonomous, battery-operated underwater gliders to cooperatively navigate and achieve desired goals
• Built and tested a simulation environment for a leader-follower strategy using optical communication technologies – specifically LED color detection using a camera and OpenCV

• Developed models, controllers, and handlers for a new wizard that is used to add/edit fields in transaction messages for updated version of existing application, IMS Explorer for Development (E4D)
• Developed refresh functionality for catalog connection and navigation tool for upcoming version of IMS E4D

• Mobile Content Solutions QA Developer: enhanced user interface, added several features, and fixed security vulnerabilities for existing in-house testing automation web application; developed an algorithm for automating input parameter generation
• Intern Hackathon: developed a suite of Alexa skills for the Amazon Echo involving various health assessments using Javascript

• Business Analyst: documented user stories and translated them into use cases that could be developed against
• Front End UI Developer: developed one of the user stories using standard web programming languages (Javascript/Angular.JS)