Shanuj Shekhar

We are developing a smart IoT climate control system and would like to leverage machine learning techniques for occupancy prediction (based on incoming sensor data) and for damper actuation (based on incoming sensor data and temperature target, when should heating/cooling be turned on and which heating/cooling dampers should be actuated).

Topic: Machine Learning

Languages and Tools Used: Python Deep Neural Networks PyTorch

I have created a dashboard for visualizing COVID-19 cases in the USA, how the disease spread and how it affected the country’s unemployment rates. I have created two modes for our dashboard - one for the unemployment data and one for the COVID-19 case data to show the spread in different states and counties.

Topic: Data Visualization, Front-end Development

Languages and Tools Used: Python D3.js HTML5 CSS3 Javascript Flask

I have used Random Forest Classifier to identify intoxicated individuals according to their TAC labels and detect drinking episodes using accelerometer samples from their mobile devices.

Topic: Machine Learning

Languages and Tools Used: Python scikit sklearn

I have designed a game in which a user can interactively build an augmented 3D scene on a planar surface in the real world. My application has two modes: a build mode and a game mode and application starts with a menu of two buttons displaying the two modes.

Topic: Virtual Reality

Languages and Tools Used: C# Unity Tool Blender 3.0 Vuforia

The task of post modifier generation requires to automatically generate a post modifier phrase describing the target entity (an entity essentially refers to a noun but here we only consider people) that contextually fits in the input sentence.

Topic: Natural Language Processing, Recurrent Neural Networks

Languages and Tools Used: Python scikit-learn numpy

This project titled “Emotion Recognition for Real-Time Feedback” performs facial expression analysis in near real-time from a live webcam feed. It classifies human expressions into 8 different classes (Happy, Sad, Angry, Contempt, Disgust, Fear, Surprise, Neutral) with an accuracy of 67% .

Topic: Machine Learning, Support Vector Machine

Languages and Tools Used: Python scikit-learn numpy OpenCV

Server-Client side scripting of Chat Application

This application allows multiple users to chat with each other. This is a very basic implemention of node.js and socket.io javascript modules. With the help of this application, users can chat irrespective of their locations.

Topic: Web Development

Languages and Tools Used: node.js nodemon ngrok HTML5 CSS3 Javascript

Performing common visual data analytic tasks using Python and D3.js.

Implemented Random Sampling and Stratified Sampling. Performed K-means Clustering for Stratified Sampling using Elbow Method and Dimension Reduction using PCA and MDS Analysis.

Topic: Data Visualization, Front-end Development

Languages and Tools Used: Python D3.js HTML5 CSS3 Javascript Flask

A numerical statistic that is intended to reflect how important a word is to a document in a collection or corpus

Calculated the term frequency for terms present in 2000 documents. (Dataset- http://mlg.ucd.ie/files/datasets/bbc-fulltext.zip). Found the top 10 most important terms for each document.

Topic: Virtual Reality

Languages and Tools Used: C# Unity Tool Blender 3.0 Vuforia

Designed Virtual Scene of Solar System consisting of Sun, Moon, Stars, cargo-ship, spaceship, etc. (Modelled using Blender 2.8).

Implemented features like gravity, switching cameras of game, revolving and rotational motion of planets, driving spaceship. Spaceship is equipped with the state-of-the-art selection ray which can pick-up and move objects using VR controllers. Designed a greenhouse on a Planet where I made the player do some random tasks like aligning cargo boxes, etc. Designed teleportation tube to teleport into the greenhouse.

Topic: Virtual Reality

Languages and Tools Used: C# Unity Tool Blender 3.0 Oculus Quest (HMD, 64GB)