MD MEHEBUB SHAIKH

• Dexapp
  

Currently working on creating a testing environment for the demand planning screen on Dexapp. Created a new page for the Truck Dispatch plan UI. Solved technical issues related to demand planning books, and helped team members to solve the issues. Documented everything and gave KT to team members.

• RAG-based Document Search Engine.

This project provides a user-friendly web interface to upload documents, ask a question, and retrieve relevant documents based on the query. It consists of an HTML/JavaScript frontend and a Python/Flask backend that simulates the retrieval part of a RAG system. It uses two foundation models, one model for vector embedding, and another model for generating a summary of the retrieved documents.

• Statistical Impact Assessment Framework Development

The objective of this study is to create an automated framework to identify key drivers for AWS customer support metrics changes between two periods. It takes metric name and period (week, month, year) as user input, and then provides key drivers and data-driven business decisions as output. For this analysis, statistical tests (parametric and non-parametric), correlation analysis, central tendency and dispersion analysis, Autogluon ML model, and language model are used.

• Escalation prediction for AWS customer support cases

It is a binary classification task to find out whether a support case needs escalation to support engineers or not. For the analysis, new features are created from the first customer and support engineer interaction using prompt engineering techniques. As the dataset is highly imbalanced, the undersampling technique is used. The Autogluon ML model is used to find the best performing model on the training dataset, and it is selected to generate a confusion matrix on the validation dataset.