Graduate Research Assistant Jan 2020 - Present
Syracuse University College of Engineering and Computer Science

Research Project 1 - Step Counting and Speed detection
Worked on a project which focuses on predicting the step counts and the person's walking speed based on the accelerometer data of the wrist & waist. Prepared the raw accelerator data into features of a specific window sizes. Fetched the data within the boundary of window sizes and converted those into tensors. Implemented CNN as baseline model and Prototypical Learning (Few Shots learning) using PyTorch. Able to achieve 96% test accuracy and 99.5% validation accuracy.

Research Project 2 - Early age stammering prediction in children
Working on a project which helps to detect/ predict stammering in children using their free speech and scripted samples. Implementing a Multiple Instance Learning attention-based model to fulfill the task.

Technologies Used: Python, PyTorch, Pandas, NumPy, Matplotlib, Scikit-learn

Mobile Application Developer Dec 2018 - Jan 2020
Indian Institute of Technology – Mumbai, India

Built a hybrid mobile application using Flutter to help people (mostly children) with severe speech or language problem. The application contains important words and enhanced illustrations to help them learn and communicate with the outside world. It contains more than 10,000 icons carefully separated in 35+ categories and subcategories. All the icons are also segregated by the their type. Built various plugins for improving the efficiency of application by approx 35%. Taking the targeted audience into consideration and their requirement, built two special keyboard serial ABC and qwerty with simplified keys. Simplified sentence construction for the children by generating live suggestion of words using natural language processing. Deployed beta version of the application which helped to identify bugs and functionality issues and later published the live version for the same on Google Play store as “Jellow Plus”

Technologies Used: Flutter, Android Studio, Kotlin, Python, Firebase Storage service

Mobile Application Developer Feb 2019 - Jul 2019
Malhar Mehta, Ahmedabad, India

Developed hybrid application for tracking and collecting information. The application was integrated into Google Maps API for tracking and accessing real-time location. The database was managed on SQL for customer details, Firebase Realtime Database for storing travel information and Firebase Storage for uploading invoices. Firebase authentication services are used to provide authentication to users. The application was published on Google Play Store as “Rudra Marketing Squad”.

Technologies Used: Flutter, Android Studio, Kotlin, Python, Excel, Firebase Storage, Authentication and Realtime-Database