Projects

Currently, I am working on a Human Activity Recognition (HAR) project for my Machine Learning class to guide simple exercise routines and gauge the accuracy of the user while performing these routines.As a proof-of-concept, I want to develop a smartphone application that utilizes a trained deep learning model for human activity recognition and serve as a personalized fitness instructor.

An application in the space of healthcare for my NLP class aimed to help doctors get a quick intuitive understanding of the patient EHR on their smartphones through concept extraction using deep learning techniques.

The platform leverages knowledge of leveled books read by children to infer their reading prowess. Additionally, the platform automatically estimates the level of unleveled books by correlatingreading patterns ofreaders inthe same cohort. The application aims at identifying books encouraging each reader to gradually progress their reading abilities and create better readers [Check it out](https://github.tamu.edu/pages/ankurrc/cs670-Blyton/)

Developed a how-am-I-driving/report-a-driver platform based on the Actions on Google framework for Google Assistant/Smart Home. This system can be used to crowd sourcing driving quality data. The collected data can be analyzed to produce a composite driving performance/safety score.

Conducted feasibility studies and cost analysis to recommend effective cloud solutions. Created an interactive Python wrapper for Docker to handle unknown dependencies. Implemented a Python CLI to emulate a simple OpenStack API to manage instances, flavors & images.

Created a Python-based UAV planning application for wilderness search and rescue ops using convolutional local hill climbing search

Built a Raspberry Pi based system with a touchscreen interface featuring a PyQT based GUI, lane departure warning system, facial recognition, rear-view camera stream, emergency SMS service and weather information.

Authentication based on not just password-content but also on biometric typing pattern. Implemented statistical clustering (k-means and neural networks (MLP) using inter-key time and key-press time as feature vectors.