Published:
Developed a FastAPI web API for real-time toxic language detection in user comments. Implements multi-label classification (toxic, severe_toxic, obscene, threat, insult, identity_hate) using TF-IDF and Logistic Regression. Features a modular, Dockerized architecture for easy deployment.
Published:
Takes one or more PDFs as input and then users can ask questions and it will answer based on those provided documents.
Published:
A fine-tuned BERT model that categorizes cybercrime complaints based on the victim, type of fraud, and other relevant parameters.
Published:
It detects 11 Bengali vowels taking handwritten figures through live sketchpad as input and predicts accordingly.