momijo is a set of native Mojo libraries. It offers fast, production-ready tools for machine learning, deep learning, AI, and data science. We focus on three things: high performance, simple APIs, and modular architecture. The goal is to be the standard Mojo toolbox for data, models, training, and deployment.

Developed a hands-on tutorial on deploying a pre-trained YOLOv3 object detection model using FastAPI.

A high-performance Deep Learning model for classifying animal faces (Cat, Dog, Wild), powered by MobileNetV2 and TensorFlow. Deployed as a fast, asynchronous REST API using FastAPI to provide real-time inference and seamless integration.

Built a deep CNN with SE blocks, attention mechanisms, and EfficientNet transfer learning for dish classification. Optimized the model with ONNX (graph optimizations + quantization) for faster inference and deployed via Flask API + Docker. Developed a Streamlit UI for real-time predictions, enabling a complete end-to-end ML solution.