1. Document Q&A RAG System FastAPI 

A powerful Document Question-Answering system built using Retrieval-Augmented Generation (RAG) architecture, FAISS Vector Search, and a FastAPI backend. It supports both Bengali and English for querying documents and provides context-aware answers. The system is also integrated with Streamlit for an interactive, user-friendly interface. 

Tech Stack: LLM, RAG, Llama 3.2, FAISS, FastAPI, Docker

Link: GitHub

2. Bangla Sentence Punctuation Restoration

A transformer-based Bangla model was used to build the sentence punctuation model. Llama 3.2 was also used to infer with non-punctuation sentence correction. FastAPI was used to prepare the API for deployment with Docker. 

Tech Stack: BanglaBERT, LLM, Llama 3.2, FastAPI 

Link: GitHub

3. Chat Bot using LLM with Gradio

The chatbot is built with Flask for the backend and uses a pre-trained model from Hugging Face for generating responses.

Tech Stack: LLM, Gen AI, Flask, Hugging Face

Link: GitHub

4. Chattogram language to standard bangla language conversion

Developed a transformer-based Seq2Seq model to convert local Chattogram language to standard Bangla. The data was processed by the sentencepice tokenizer.

Tech Stack: Python, PyTorch, YOLOv10, EasyOCR

Link: GitHub

5. Named Entity Recognition using SpaCy transformers

Developed a transformer-based NER model to extract the entity of a document. The model was trained on the resume dataset.

Tech Stack: Python, PyTorch, Transformers, RoBERTa

Link: GitHub

6.  Bangla Roberta Question and Answer

Bangla Question Answering model architecture is the BERT-based Roberta Model, which is trained on Bangla QA data. For training this model, the Bangla QA data is converted into the SQuAD v2 format.

Tech Stack:Python, PyTorch, Transformer

Link: GitHub

7. Resume Categorization using Transformer Model

The aim of the project is to categorize resumes using a transformer-based document classification model. It involves preprocessing resume data, training a BERT-based model, and creating a Python script that classifies PDF resumes into predefined categories.

Tech Stack:Python, PyTorch, Transformer, BERT

Link: GitHub

1. Automatic Wrong Side Vehicle Detection System

Developed a YOLO v10-based wrong-side vehicle movement detection. The wrong side vehicle was detected using a custom-trained YOLO v10 model and the license plate number was extracted using EasyOCR.

Tech Stack: Python, PyTorch, YOLOv10, EasyOCR

Link: GitHub

2. Image Cryptography with Autoencoders 

Image encryption and decryption using a chaotic map sequence and an autoencoder. It includes code for generating chaotic map sequences, shuffling and deshuffling images, preparing datasets, and computing performance metrics.

Tech Stack: Python, TensorFlow, OpenCV, CNN, Security

Link: GitHub

1. Telco customer churn prediction using machine learning

In many service-oriented industries, retaining customers is as crucial as acquiring new ones. Customer churn, which refers to customers ceasing their relationship with a company, is a significant metric that companies track to understand their service performance. 

Tech Stack: Python, EDA, Scikit-learn

Link: GitHub

1. Tech World Website

Technological information-related news website. One can create a technological post. There is an admin panel who can manage all the posts on his website. 

Tech Stack: HTML, CSS, JavaScript, Bootstrap, PHP 

Link: GitHub

2. Game Contest Management System

The aim of this project is to launch a gaming competition in KUET. The purpose of the project is to properly maintain the players' information, volunteers' information, game categories management, and individual ranking system. 

Tools: Javas, Laravel

Link: GitHub