Shimaa Ahmed

Shimaa Ahmed

PhD Student

University of Wisconsin–Madison

About Me

Open to New Opportunities: I am currently on the job market seeking roles in Applied ML/AI, Responsible AI, Speech, and Privacy in AI, in industrial and academic research labs. My research experience includes work on speech-related machine learning models such as speech recognition, keyword spotting, voice cloning, and speaker identification, and generative AI for image generation and manipulation for face recognition applications, and LLMs governance. I also have experience on privacy-preserving techniques such as differential privacy and training on private data.

I am a Ph.D. dissertator in the Electrical and Computer Engineering Department at the University of Wisconsin–Madison, under the supervision of Prof. Kassem Fawaz at WI-PI lab. My research interests lie at the intersection of Security & Privacy, Machine Learning, Systems, and fair and trustworthy artificial intelligence. Through my research, I examine the risks posed by ML algorithms and develop robust systems to mitigate these risks and empower users with agency over their personal data. Specifically, I have explored different ubiquitous ML applications encompassing various data modalities such as speech, text, and vision. Check my research statement for more details.

In the past, I made research contributions to Massive MIMO for wireless communications as well as Analog and RF IC design.

Interests
  • Trustworthy Machine Learning
  • Speech Recognition
  • Conversational AI
  • Large Language Models
  • Image Synthesis and Manipulation
Education
  • PhD in Electrical and Computer Engineering, 2017–Present

    University of Wisconsin–Madison

  • MS in Electrical and Computer Engineering, 2014–2017

    Ain Shams University

  • BS in Electrical and Computer Engineering, 2007–2012

    Ain Shams University

Recent News

Sept 2023: I have been selected for EECS Rising Stars, Georgia Tech.

Sept 2023: News coverage of our paper Tubes Among US: Analog Attack on Automatic Speaker Identification on Newspaper, Online platforms, NPR Radio, Podcast, News 3 Now TV (CBS affiliate), UW–Madison Campus News, UW–Madison social media

Aug 2023: Presented our work Tubes Among Us at USENIX Security'23

May 2023: I will be teaching ECE697: Capstone Project in Machine Learning and Signal Processing for the Professional Master program at UW–Madison.

Feb 2023: Notable Reviewer recognition at 1st IEEE Conference on Secure and Trustworthy Machine Learning (SatML)

Sept 2022: Presented at the DARPA GARD (Guaranteeing AI Robustness Against Deception) program


Experience

 
 
 
 
 
Wisconsin Privacy and Security Group
Research Assistant
Sep 2018 – Present Madison, WI
I design privacy-preserving technologies for ML applications to enhance their robustness and integrity, with a focus on speech-related technologies.
 
 
 
 
 
University of Wisconsin-Madison
Graduate Student Instructor
May 2023 – Aug 2023 Madison, Wisconsin

Course: ECE 697: Capstone Project in Machine Learning and Signal Processing

  • Delivered lectures on topics related to ML tools and applications
  • Mentored 8 groups to do projects in cutting edge topics in AI and ML.
  • Invited more than 15 speakers from industry and academia
 
 
 
 
 
Cleverhans Lab at University of Toronto and Vector Institute
Visiting Scholar
Jun 2022 – Sep 2022 Toronto, Canada
worked on implementing robust and trustworthy ML models by analyzing the model’s training dynamics under the supervision of Prof. Nicolas Papernot.
 
 
 
 
 
Microsoft, AI Platform
Research Intern
Jun 2021 – Sep 2021 Redmond, WA

Robust Training Techniques for Keyword Spotting, under the supervision of Dr. Anthony Stark and Dr. Jian Wu

  • implemented multiple data selection and re-weighting techniques to efficiently train keyword spotting models.
  • utilized both natural and synthetic data and designed an alogorithm that weigh each sample baed on its acoustic sequence and model sensitivity.
  • applied these techniques to customized keyword spotting models, where training data is sparse, and commercial models, where training data is abundant but not equally significant to the model’s learning.
 
 
 
 
 
Si-Ware Systems, Timing Solutions Division
Analog IC Design Engineer
Sep 2012 – Feb 2014 Cairo, Egypt
designed an All-Digital Phase Locked Loop that is driven by an All-silicon reference oscillator.
google-site-verification: googlef11aad333c6a90a4.html