Motivation

BirdML began as a passion project built around my love for birds. Birds are one of the most diverse and ecologically important groups of animals on Earth. Beyond their beauty, they play important roles in seed dispersal, pollination, pest control, disease prevention, and maintaining healthy ecosystems.

At the same time, many bird species are facing serious threats from habitat loss, climate change, and other environmental pressures. By helping users identify bird species from images, this project aims to make birds feel more visible, familiar, and memorable. My hope is that a simple prediction can spark curiosity, encourage people to learn more about the species around them, and raise awareness about the importance of protecting biodiversity.

MLOps and Technical Purpose

This project was also built to showcase my ability to develop and productionize an end-to-end machine learning system. The application uses an EfficientNetB3 convolutional neural network trained with transfer learning to classify bird images across 525 species. The model was trained using the yashikota/birds-525-species-image-classification dataset, which contains 89,885 bird images.

The system includes a backend prediction API that loads the trained model and serves inference requests, as well as a frontend that allows users to upload bird images and receive predictions. The training pipeline was developed using AWS SageMaker, with the goal of creating infrastructure that can support future model improvements, additional bird species, and eventually a feedback system where users can rate prediction quality.

Images uploaded by users are not collected on this site.

About Me

Hi, I'm Hamad Alajeel. I have a bachelor's and master's degree in Electrical Engineering from UC San Diego, with a focus on Machine Learning and Data Science. I am passionate about ML engineering and building systems that turn machine learning models into real, usable applications.

During my academic work, I built projects involving recommendation systems, computer vision, generative models, and diffusion models. However, most of that work was completed in Jupyter and Google Colab notebooks, which did not fully reflect the production reality of machine learning systems.

This project helped me bridge that gap. It gave me the opportunity to work beyond model training and focus on the full ML lifecycle, including data handling, training pipelines, backend inference, frontend integration, cloud deployment, and future system monitoring. My goal is to continue building practical AI and ML systems that are reliable, useful, and impactful.

Contact

For questions, feedback, or professional opportunities, feel free to reach out through