Ali Elganzory

Green AI Engineering

• Transformed market requirements into new features and functionalities using React and Laravel. • Put in place anti-AI-cheating mechanisms for written responses to flag suspicious activities. • Implemented and deployed candidate recommendation system to production. Demoed it to market experts as a part of a 1-week hackathon and secured 1 st place.

• Initiated the idea of forecasting mobile network traffic to anticipate any over-capacity needs in infrastructure. • Feature-engineered external real-world factors – i.e., soccer matches of teams and leagues in and out of Egypt. • Improved the baseline Long Short-Term Memory model's error (Root Mean Square Error) from nearly 31.83 to 14.63 using a Multilayer Perceptron. Constructed models using PyTorch and Sklearn.

• Explored, preprocessed, analyzed, and visualized a simulated dataset of 20 types of physical particles. • Comparatively experimented with XGBoost, Deep Neural Networks, and K-Means algorithms for predicting and clustering particle types. • Differentiated particles from anti-particles with an accuracy of 98.11% and predicted specific particle types with an accuracy of 88%.

• Utilized Light Detection and Ranging (LiDAR) for the detection of cones in the 3D space around the car. • Optimized the performance of the LiDAR module and decreased its inference time 3-fold by vectorizing the pipeline of filtering, clustering, and classification of road cones using Python’s PCL and NumPy. • Designed a supervisor module using ROS and Python, which bootstraps and monitors the system, and interfaces with the vehicle’s control.

Exploratory and Explanatory Data Analysis, Practical Statistics, A/B Testing, Data visualization, and Presentation.

• Cumulative GPA: 3.72 with honors. • Specialization: Data Science. • Thesis: Semantic Manipulation of Objects in Images.