Hey there π, I'm Hailemariam Mersha
I am currently a penultimate-year Computer Science and Mathematics student at NYU Abu Dhabi and a Student Research Assistant at the eBrain Lab and the SANAD Lab.
I'm passionate about AI security, computer vision, deep learning, and applied machine learning research. I have a strong drive for exploring cutting-edge technologies and solving complex problems through first-principle thinking.
When I'm not coding or conducting research, you'll find me organizing cultural events, leading meditation sessions, or watching tech-related content to fuel my curiosity.
Currently, I'm focused on:
- LLM Security & Robustness β Investigating jailbreak techniques, vulnerabilities, and defense mechanisms in large language models, detecting and Measuring Prompt Injection Attacks in Cryptocurrency Token Metadata Targeting AI Trading Systems
- Climate Data Analysis β Analyzing large-scale climate datasets (CMIP6 & ERA5) to study temperature trends over the Arabian Peninsula
- Computer Vision β Exploring object detection, tracking, and transformer-based architectures like DETR
- Deep Learning β Building neural networks from scratch and fine-tuning state-of-the-art models
- I love exploring new technologies and building innovative solutions
- Passionate about AI security, computer vision, and climate science
- Continuous learner β always diving deep into new research papers and technologies
- Community builder β organizing cultural events and creating inclusive spaces
- LLM Jailbreak Security Research β Investigating security vulnerabilities and defense mechanisms for safer AI systems
- Climate Data Analysis using CMIP6 & ERA5 β Analyzing large-scale climate datasets using Python, xarray, and HPC systems
- Bipartite Object Matching on VIRAT Frame Pairs β Implementing bipartite matching algorithms for object tracking across video frames
- CNN from Scratch & ResNet Fine-Tuning β Built CNNs from first principles with custom backpropagation
- Moving Object Detection using DETR β Transformer-based object detection in video sequences
- SwapBay β Full-stack item trading platform with authentication and messaging
- Locify β Location-based application developed at Princeton Hackathon
Languages: Python, C/C++, Javascript, HTML, CSS
Tools: Git, Docker, VS Code, PyCharm, HPC environments
Libraries: PyTorch, pandas, NumPy, Matplotlib, xarray
- π§ Email: hbm9834@nyu.edu
- πΌ LinkedIn: linkedin.com/in/hailemariam-mersha)
- π GitHub: github.com/HailemariamMersha
- π Portfolio: HailemariamMersha.github.io/portfolio
"The only way to do great work is to love what you do." - Steve Jobs

