Muneeb Elahi Malik

University of Georgia - Athens, GA, USA

Ph.D. in Engineering

Research Focus: Robotics and Applied AI in Precision Ag

Jan 2024 - Present

My research focuses on integrating robotics and applied AI in precision ag. I am working on the development of an advanced Autonomous Laser Weeder Robot.

National University of Sciences and Technology - H-12 ISB

Bachelor of Science in Electrical Engineering

Minor: Computer Science

Sep 2019 - Jun 2023

Relevant Coursework

• Embedded Systems Design
• Robotics
• Industrial Process Control
• Machine Learning
• Control Systems
• Digital Signal Processing

FYP: Forest Damage Detection using Unsupervised Learning

Designed a novel approach for forest damage assessment in data-scarce regions like Pakistan leveraging clustering techniques on multi-spectral satellite imagery. Developed a Web Portal for Pakistan's Forestry Department to accurately monitor wildfires, enabling better conservation and disaster management.

South Dakota State University - Brookings, SD, USA

Global UGRAD Exchange Program Spring 2022 - USA

Jan 2022 - May 2022

Relevant Courses

• Systems Engineering
• Leadership
• Engineering Management
• American Government

Progam: Sponsored by US Department of State and administered by IREX

Development Projects

• Audio Processing and Wi-Fi Alert Transmission using ESP32 and FreeRTOS with Microphone Interface.

  Developed a safety-enhancing system that integrates embedded systems, audio processing, machine learning, and Wi-Fi communications. The system utilizes ESP-32 and FreeRTOS to enable real-time audio analysis and transmission of classified audio signals as alerts or notifications over a Wi-Fi network.
  Tech Stack & Hardware: FREE RTOS, ESP-32, INMP441 I2S Mic, Arduino IDE
  GitHub Link

• Car Drowsiness detector – Real-time Car Safety System
  

  The Car Drowsiness Detector is an innovative solution that helps prevent accidents caused by drowsy driving. By using computer vision and deep learning techniques, the system monitors the driver's eyes in real-time and alerts them if they start to close or show signs of drowsiness.
  Tech Stack & Libraries : OpenCV, Deep Learning, Pygame
  GitHub Link

• HemoDetect: Empowering Precision Medicine through Multi-Class Disease Classification of Blood Cell Images

  Developed HemoDetect, an innovative deep-learning solution designed for accurate disease diagnosis and classification. By leveraging Faster R-CNN, the model effectively performs multi-class disease classification on peripheral blood cell images. This enables rapid and precise diagnosis of life-threatening conditions such as Malaria, Leukemia, and Sickle Cell Anemia
  Tech Stack & Libraries : Python, Pytorch, ResNET-50, TensorFlow
  Github Link

• Transformer Augmented Mask RCNN for Enhanced Instance Segmentation

  Designed an instance segmentation model by augmenting multi-head attention in Mask RCNN structure. Deployed the model for apple disease detection in complex environments.
  Tech Stack & Libraries : Python, Pytorch, ResNET-50, TensorFlow
  Github Link:

• Soft Watermarking for Large Language Models

  Implemented Soft Watermarking approach on Bart pre-trained on CNN Daily Mail dataset. Designed watermark detector, analyzed performance in low-entropy text and strong watermarking etc.
  Github Link:

Research Projects

• Forest Damage Detection using Unsupervised Learning

  Designed a novel approach using Cascaded K-Means clustering to identify anomalies in Sentinel-2 multispectral data, uncovering forest fire damage patterns. Leveraging a supervised classifier trained on spectral indices, the clusters were classified into burn severity classes. To enhance monitoring capabilities, a user-friendly Web App was developed, empowering the Forest Department to track and analyze Northern Punjab Forest Fires over the past 5 years.
  Tech Stack & Libraries : Javascript, Google Earth Engine, Sentinel-2, Weka
  GitHub Link | Paper (Coming Soon)

• Hardware Accelerator for Transformer Model

  Implemented Hardware Accelerator for Multi-Head Attention and Position-Wise Feed-Forward of the Transformer model.
  Tech Stack & Libraries : Verilog, , OpenAI
  GitHub Link | Paper (Coming Soon)