About Me
Versatile honors graduate from NMIMS University, currently pursuing masters in computer science at New York University. Experienced in developing and testing prediction models, while also visualizing and explaining the results via interactive dashboards. Looking to solve business problems using data-driven models which would help the management make informed decisions.
Technical Skills
Programming Languages :
R , Python , Java , SQL , C++ , HTML , CSS , Hadoop, Spark , GIT
Web Development :
Bootstrap , JavaScript , D3.js , Flask , Dash , Streamlit
Machine Learning Technologies :
Regression (Ridge and Lasso, Decision Trees , Bagging , Boosting , Random Forests , SVM , KNN , K-Means , PCA , t-SNE , Neural Networks (CNN,RNN,LSTM ,GAN), BERT , RoBERTa , BLOOM , SHAP, LIME
Libraries and API:
Pandas , Numpy , Scipy , OpenCV , Scikit-learn , Matplotlib , Seaborn , Altair , Beautiful Soap, Plotly , Tweepy , nltk , dplyr , ggplot, caret , tidyr , TensorFlow , PyTorch , Keras , Dask , IBM Watson , Discord
Other Skills :
MATLAB , Microsoft Excel , SAS , StarUML , Android Studio , Flutter , AWS , Google Colab , ETL , Regression Analysis, Docker , MongoDB , EDA , Google Analytics
Projects
- Developed an open domain DialoGPT based chatbot fine-tuned on a multilingual conversation data in Hindi and English
- The bot is scalable, can chat with multiple users concurrently while maintaining separate chat history for customized replies
- Deployed the discord bot on Heroku and used Accelerated Inference API for generating responses with perplexity of 3.21
- Led a team to deploy a real-time analytics dashboard which predicts the current mental state and displays the user history
- Hosted the streamlit application on AWS as containers using ECS and Load Balancer which was secured using AWS Cognito
- Built a pipeline that takes user input from front-end and generates the predicted output in less than 1 ms
- Developed a RoBERTa and SpanBERT based question answering models (Available on Huggingface) to extract the emotion cause in each statement
- Performed exploratory data analysis in python to improve the explanation model by eliminating issues like class imbalance
- Presented the models as an alternative to SHAP and was proven better statistically using metrics like IoU
- Developed a chatbot which recommends restaurant based on user inputs like preferred cuisine and locality
- Collected restaurant data from Yelp API and stored it in DynamoDB and Open Search (Elastic Search)
- The recommendation system sends an E-Mail using SNS suggesting 5 restaurants which satisfy the user preferences
- Developed an application that uses Text and Audio/Video to analyze a person’s personality by applying transfer learning
- Tested the model on Essays dataset and achieved the best accuracy till date along with real-world data collected via survey
- Prepared a Dashboard to visualize the prediction and present analysis by describing each chart to explain the AI results
- Detected and Classified faces into 7 different categories like Angry, Sad, Fear etc using CNN and OpenCV
- Developed a webpage using Flask to use webcam feed as input and generate live output serving the prediction of the emotion in real time on the web interface
Experience
Dhandhania Infotech
Data Intern
March 2020 - March 2021
Incorporated Twitter and cover letter data for personality analysis of job applicants to aid human resources department in elimination of unsuitable candidates based on personality traits
Education
New York University, Tandon School of Engineering, New York
Masters of Science, Computer Science
2021 - 2023
Relevant Courses
Machine Learning, Computer Vision, Big Data, Cloud Computing, Visualization , Data Science for Business
Narsee Monjee Institute of Management Studies (NMIMS)
BTech (Hons), Computer Engineering
2017 - 2021
Relevant Courses
Probability and Statistics , Business Visualization , Human Resource Management , Data Mining , Predictive Modelling , Artificial Intelligence , Operations Research , Image Processing , and Database Management Systems