Gregory Ofoegbu

Cincinno - Facial Recognition Deadbolt

• Developed the frontend and backend using .NET Core, creating a responsive and interactive user interface along with a robust backend API.
• Implemented a RESTful API to handle user authentication, data processing, and interactions between the client-side application, the server and a Raspberry Pi.
• Designed and managed a PostgreSQL database, ensuring efficient data storage, retrieval, and optimization.
• Deployed the application on AWS, leveraging scalable infrastructure to support multiple users uploading pictures to train a facial recognition model.

Audio Classifier - Machine Learning

• Developed a machine learning model to classify marine mammal species based on their acoustic signals, achieving an accuracy of 92%.
• Utilized Python and libraries such as TensorFlow/Keras for model development, leveraging deep learning techniques like Convolutional Neural Networks (CNNs) for feature extraction and classification.
• Preprocessed audio data by applying techniques such as Fast Fourier Transform (FFT).
• Documented and visualized model performance using confusion matrices providing insights for further refinement.