Khurram Khalil

Khurram Khalil

PhD Candidate - AI Model Security & Optimization (DNN & Generative AI)

About Me

I am Khurram Khalil, a PhD candidate in Computer Science at the University of Missouri-Columbia, specializing in AI Model Security & Optimization with a focus on Deep Neural Networks (DNN) and Generative AI. My research interests encompass machine learning, computer vision, and cybersecurity, with a particular emphasis on developing robust and efficient AI models.

Research Focus

My doctoral research concentrates on enhancing the security and efficiency of AI models, with particular emphasis on:

  • Adversarial machine learning and defense mechanisms
  • Optimization techniques for deep neural networks
  • Applications of generative AI in security contexts
  • Explainable AI and model interpretability
  • Multi-modal learning and fusion techniques

Education

  • Ph.D. in Computer Science, University of Missouri-Columbia (Expected 2028)
  • M.S. in Robotics and Intelligent Machine Engineering, National University of Sciences and Technology (NUST), Pakistan, 2021
  • B.S. in Electronics Engineering, International Islamic University (IIU), Pakistan, 2018

Professional Experience

  • Graduate Research Assistant, University of Missouri-Columbia (2020-Present)
    • Conducting research on adversarial machine learning, model robustness, and AI security
    • Developing novel techniques for multi-modal learning and fusion
    • Collaborating on projects involving autonomous systems and robotics

  • Sr. Machine Learning Engineer
    AxcelerateAI | Dec 2023 – Aug 2024
    Focused on diffusion models, Stable Diffusion, image inpainting, OpenAI APIs, and LLMs, leading AI strategies and implementations in Digital Healthcare, Fintech, Retail AI, and Smart City Solutions. Expertise includes guiding a skilled team in AI-driven transformation, utilizing GCP, MLOps, and cutting-edge 3D image editing.

  • Sr. Machine Learning Engineer
    Center for Computing and Artificial Intelligence (CENTAIC), PAF | Nov 2020 – Dec 2023
    Led projects in pattern analysis, spatiotemporal tracking, explainable AI, and multi-sensor tracking. Key focus on deploying AI algorithms for advanced scientific data visualization and complex problem-solving in real-world AI applications.

  • Artificial Intelligence Engineer
    Arbisot, Westwood, Lahore (Remote) | Jun 2022 – Sep 2022
    Specialized in constraint-based optimization, time series analysis, and machine learning for electric vehicle fleet modeling, utilizing Bass diffusion models and dynamic optimization.

  • Deep Learning Engineer
    Spotmydot AB, Massachusetts, USA (Remote)
    Developed custom auto-grad implementations, worked on image recognition and optimization algorithms like PSO and GA for real-world AI systems.

Selected Publications

  • Khalil, K., Asgher, U., Ali, S., et al. (May 2023). Time and accuracy based comparative analysis between machine learning and deep learning algorithms: an fNIRS study. Artificial Intelligence in Medicine. (Under review)
  • Mughal NE, Khan MJ, Khalil K, et al. (2022). EEG-fNIRS-based hybrid image construction and classification using CNN-LSTM. Frontiers in Neurorobotics. https://doi.org/10.3389/fnbot.2022.873239
  • Khalil, K., Asgher, U., & Ayaz, Y. (2022). Novel fNIRS study on homogeneous symmetric feature-based transfer learning for brain-computer interface. Nature’s Scientific Reports. https://doi.org/10.1038/s41598-022-06805-4
  • Full Publication List

For a complete list of my publications, please visit my Google Scholar profile.

Skills

  • Programming: Python, MATLAB, C++, ROS
  • Machine Learning Frameworks: TensorFlow, PyTorch, Keras
  • Computer Vision: OpenCV, PCL
  • Data Analysis and Visualization
  • High-Performance Computing
  • Technical Writing and Presentation
  • Embedded Systems and Microcontrollers

Selected Projects

Temporal Pattern Analysis | Oct. 2022 – Present

Real-time analysis of time-based events using NumPy, Pandas, shapely, and rtree.
GitHub

Multi-object Tracking (MOT) | Dec. 2021 – Jun. 2022

Developed multi-sensor tracking systems for targets that are neither fully nor directly observable. Tools: NumPy, networkx, plotly.
GitHub

Spatio-temporal Constraint Optimization | Jun. 2021 – Jul. 2022

Non-linear dynamic optimization using ortools, gekko, and NumPy for real-time decision-making.
GitHub

For more information about my research and projects, please explore my GitHub repositories or contact me directly at kkhalil.pg@smme.edu.pk.

I am always open to collaborations and discussions on cutting-edge research in AI, machine learning, and robotics. Feel free to reach out if you’re interested in my work or potential collaborations.