Natan Mesay Email & Phone Number

Addis Ababa, Et

Addis Ababa, Ethiopia

In the second phase of my internship, I gained valuable experience in machine learning and deep learning. I worked on predictive analytics projects utilizing supervised learning techniques. First, I built models to predict customer churn by analyzing demographic and transactional data, helping the company better understand customer retention risks. I also created models to forecast car prices based on vehicle attributes and metrics. This allowed dealerships to more accurately value used cars on their lots. During these projects, I applied concepts like overfitting, underfitting, and testing models for accuracy. The technologies I used were Python, Jupyter Notebook, and MATLAB for data visualization. This exposed me to neural networks and the machine learning lifecycle from data collection to model deployment. I learned to preprocess data, select appropriate algorithms, tune hyperparameters, and evaluate results. Challenges like dealing with imbalanced datasets gave me opportunities to research techniques like resampling. Presenting my findings uncovered additional areas for improvement.

During my internship, I gained valuable full stack development skills working with Spring Boot, Java, REST APIs, React, and Docker. I developed backend APIs for an internal tools application, securing endpoints and safely handling sensitive employee data. This gave me hands-on experience with Spring annotations that streamlined my codebase by 20%. I also integrated a React frontend application to access the APIs, completing my full stack development knowledge. Applying best practices, I secured the APIs and data to ensure compliance. Further, I packaged the application using Docker containers, improving scalability and portability. Through this work, I learned how to deploy APIs, apply authentication and authorization, encrypt sensitive information, and filter requests - all vital skills for web development. It was a challenge to integrate the new frontend while maintaining the established Java backend, but I overcame this by dedicating extra time to learn React integration patterns. Similarly, managing dependencies across containers required extensive research and testing, which I solved through testing multiple deployment configurations.