Phish-Hawk V2

This project is the next version of Phish-Hawk, upgrading its core logic from a simple keyword-checker to a smart AI model. While users still upload files or paste text just like in V1, the analysis is now far more intelligent. The biggest change was replacing the old, rigid keyword system that caused false positives. I built a custom deep learning model in Python using PyTorch, training it on over 190,000 real emails. This new "brain" can now understand the context of an email, making it much better at spotting real phishing patterns while ignoring legitimate messages. To make this work, I re-architected the project into two parts: the main Java/Spring Boot application and a separate Python microservice (using FastAPI) that serves the AI model. The Java app calls the Python service to get an analysis, showing how different technologies can work together in a modern application. The result is a more accurate and powerful Phish-Hawk that fixes the main limitation of the original version.