Chetna Khanna
Master of Science - Data Analytics Engineering, Northeastern University
I am an educationist (with over 5+ years of experience in EduTech domain) who is transitioning into tech as a data professional. I am in love with the field of data science and analytics. Reason - Data has a way to surprise me always with the patterns I could not even think of and there comes the role of science to find the unknown. Being a Mathematics student, I have always loved finding the unknowns (x 😉 ). The “why” associated with these patterns helps create a connection between my analytical and creative self and helps me gripped with my work.
I consider myself a problem solver and a forever learner. 📘 I love to read, I love to write and I love to share my knowledge.
To provide me a better understanding of how patterns are found from data and decisions are made, I decided to extend my previous master's in mathematics to data science. I am a proud husky of Northeastern University. 🎓
Experience
NLP Engineer Co-op
Jeevahealth is a health, wellness, and fitness startup using artificial intelligence to build mental health solutions for school and college-level students
- Open Source Project: Designed DialogFlow conversion process to simplify chatbot building using one-line terminal command that provides full support - input & output context setting, parameter saving, lifespan setting, entity tagging
- Built command-line utilities to extract data from online forums like Reddit, Discord & counseling websites
- Working on building a conversational AI using RNN & deploying using Flask, Gunicorn & Docker on AWS EC2 instance
- Performed data cleaning on corpora extracted from various online forums & counseling websites
- Coordinated with team members across the globe for server setup, local installations & pipeline setup
- Did market research of health & wellness companies already present & the solutions they provide
Machine Learning Engineer Intern
Developed an end-to-end AI-enabled customer support chatbot as a service integrated with slack & deployed on cloud
[Github]
- Scrapped & processed the Nutanix public documentation to build knowledge base in JSON format
- Used keywords extraction & LSTM neural network model to predict responses & related article headings & provided links along with some portion of article content
- Chatbot listens to customer queries in passive mode, answers customer queries that can be answered with high confidence/accuracy (Responds to the query if the prediction probability surpasses a threshold value)
- Customers can also explicitly wake the chatbot using the word "@missionx"
- Uses thread to organize discussions
- Chatbot will be rolled out to 14000+ enterprise customers
Subject Matter Expert
- Designed Profit & Loss statements to help in commisioning profitable educational products
- Performed weekly analysis of per-user data of Ask and Answer forum (with queries of over 6.5 million users) & ensured answer submission within 24 hours leading to an increase in user registration by almost 10%
- Researched international curriculums (leading websites, publishers and books) to gain insight into the business model
- Understood stakeholder's requirements & worked with cross-functional teams to assimilate the requirements
- Conducted product presentations at the school and national sales level
Education
Northeastern University
- Deep Learning and Neural Networks
- Natural Language Processing
- Data Mining in Engineering
- Data Management and Database Design
- Probability and Statistics for Engineering
- Computation and Visualization for Analytics
- Special Topics in Artificial Intelligence
Hindu College, University of Delhi
Probability Theory, Statistics, Linear Algebra, Complex Analysis, Optimization Techniques & Control Theory, Mathematical Programming, Operations Research
Ramjas College, University of Delhi
Probability & Mathematical Statistics, Algebra, Analysis, Linear Programming & Game Theory, Numerical Analysis & Computer Programming, Calculus
Projects
CIFAR-100 Object Recognition
CIFAR-10 Object Recognition Using Convolutional Neural Networks [Github]
- Built a CNN model from scratch using TensorFlow to recognize and classify colored images into one of the 100 classes & achieved an accuracy of 59% using image augmentation, max pooling, zero padding, dropout, early stopping
- Trained a CNN model using current state-of-the-art EfficientNetB0 with image net weights & achieved accuracy of 82% using image augmentation, early stopping, reduce learning rate on plateau & hyper parameter tuning techniques
- Trained the transfer learning model on NVIDIA GPU and 8 vCPUs machine provided by Google Cloud Platform
Text Summarization
CNN/Daily Mail, Media Dataset Summarization using Long Short Term Memory
- Designed a neural attentional sequence-to-sequence LSTM model that can generate abstractive summaries of text
Predict Member Churn
Leading US Wholesale Corporation, Customer Dataset
- Processed 10 GB industry data to predict member churn using Logistic Regression, k-NN, Naive Bayes, Decision Tree & Random Forest
- Using model selection - cross-validation technique achieved highest accuracy of 72.2% from Random Forest model
COVID-19 MONITOR
Developed a COVID-19 dashboard built using R Shiny to monitor geospatial trends of COVID-19 across the globe [App]
- Monitor displayed live data to showcase the total confirmed, deaths, active & recovered cases using animated & non-animated visualizations
- Data was fetched through an automated ETL pipeline connected to John Hopkins COVID-19 data repository on GitHub
Property Management System
Developed database to store information related to rental services and maintenance for residential property, residents, and staff as a one-stop solution [Github]
- Developed database schema, tables, table-level check constraints, computed column & encryption using SQL server
- Created views to know the apartment portfolio availability, pet owners information, upcoming expiring leases & pending maintenance requests
- Fetched data from the database to present business reports using Tableau
Oyewiki (Blogging Website)
Developed a DIY tool for authors to write & publish articles, & earn per view (YouTube for readers & writers)
- Extracted data from MySQL database stored in AWS S3 as SQL files using Python script to perform data analysis [Github]
- Analyzed data of 10,000 users, 22,000 articles, millions of views to find top read articles, categories & users
- Did marketing via Quora, email, text messages to increase the website traffic
- Tracked the fake views generated on articles by users
Skills
- Language: Python, R, SQL, JAVA (Basic), HTML (Basic)
- Library: Pandas, NumPy, Scikit-Learn, Matplotlib, Seaborn, TensorFlow, Keras, BeautifulSoup, NLTK, spaCy, Flask, R Shiny, ggplot2, dplyr, tidyverse
- Database: SQL Server, PostgreSQL, MySQL
- Software/Tool: Git, Github, Gitlab, GCP, AWS (Basic), Tableau, Jupyter Notebook, Colab, Excel, Docaker, PowerPoint, Jira, PyCharm
- Algorithm: Regression, Classification, Clustering, Machine Learning, Deep Learning / Neural Networks
- Statistical Methods: Hypothesis Testing, Confidence Interval
Interests
Apart from being a technology enthusiast, I enjoy going on long drives with my family 🚘. Traveling is my hobby and its fun, adventure, excitement, and happiness for me 😇. I love to capture the beauty of nature through my photography 📷. I love to meet new people 👋🏻 and explore the beauty of this world 🏝 🗻 🌈 .
I love to cook 🍽 🌮 and anything related to art and craft excites me a lot 🎨.
Awards
- Dean’s Seattle Leaders & Scholars Award, Northeastern University, 2019
- Ace Award (Best employee award), Vikas Publishing, 2018
- Star Performer, Vikas Publishing, 2017
- Steve Jobs (Best creative idea), Pearson Education, 2016
- Pinnacle (Best Math series), Pearson Education, 2016
- Bright Spot, Pearson Education, 2015
- Sitaram Jindal Foundation Scholarship, 2009-11
- Scholarship for Class X achievements, CBSE and DAV School, 2005
- Laureate Certificate and Certificate of merit (100% marks in Math), 2005
- Pratibha Sammaan (More than 85% marks in Class X), 2005