Benchmarking Classical Machine Learning and Deep Learning Transformers for Bangla Hate-Speech Detection. ● Designed an experimental workflow comparing classical ML baselines with deep learning transformer models. ● Worked with the BANHATE dataset and focused on consistent preprocessing, stratified evaluation and comparative model analysis. ● Conducted a comparative study of classical machine learning and transformer-based approaches for Bangla hate-speech classification. ● Analyzed classification performance using standard evaluation metrics including Accuracy and Macro-F1. ● Focused on model comparison and evaluation for a low-resource Bangla NLP classification problem. ● Research area: Bangla NLP, hate-speech detection, text classification and transformer-based deep learning. ● Documented methodology and experimental results as part of the undergraduate thesis.

AI Chatbot Model. ● Built an AI chatbot application using LangChain and structured runnable components including RunnableBranch and RunnableParallel. ● Used Pydantic-based validation and Streamlit to create an interactive application interface. ● Designed the project to demonstrate practical LLM orchestration and modular AI application workflows.

Mobile Price Prediction. ● Developed a machine learning application for predicting mobile phone price categories from technical specifications. ● Applied preprocessing and feature engineering to prepare structured input data for model training and inference. ● Deployed an interactive prediction demo on Hugging Face Spaces.

FastAPI Projects. ● Built a REST API for student-related data operations with FastAPI. ● Implemented API routing, request validation and structured response handling. ● Developed a backend Todo application demonstrating CRUD-oriented API workflows. ● Implemented backend routes and data-handling logic using FastAPI.