Shivani R Rapole
Education
- Master of Science in Engineering in Computer and Information Science (CIS/MSE), University of Pennsylvania, Concentration: Artificial Intelligence, 2021-2023 [current GPA: 3.9/4.0]
- M.Sc. (Hons.) Economics and B.E. (Hons.) Computer Science, Birla Institute of Technology and Science (BITS) Pilani - India, 2015-2020 [GPA: 9.4/10]
Work experience
- September 2021 - Present: Graduate Research Assistant
- Location: World Well-Being Project, University of Pennsylvania
- Duties included:
- Leveraging single- and multi-task Natural Language Processing to predict depression symptoms and languageanalysis from around 1 million Reddit posts
- Supervisor: Professor Lyle Ungar and Prof. Sharath Chandra Guntuku
- June 2020 - July 2021: Graduate Research Assistant
- Location: Computational Data Science Department, Indian Institute of Science, Bangalore - India (Collaboration with the Royal Dutch Shell Technology Centre, India)
- Duties included:
- Developed a GAN (Generative Adversarial Network)-based deep learning technique to successfully generate super-resolved digital rock images from their low-resolution digital rock images which were acquired using CT scans of various sedimentary rocks. Studying the porosity of these super resolved rocks will facilitate the oil companies to identify the suitable location for oil drilling.Tech Stack: PyTorch library in Python, CUDA.
- Worked on developing a 3D CNN model to simultaneously super-resolve and segment the low-resolution digital rock images. Tech Stack: PyTorch library in Python, CUDA.
- Supervisor: Professor Phaneendra Yalavarthy
- June 2019 - April 2020: Undergraduate Research Intern
- Location: Indian Institute of Science, Bangalore - India (Collaboration with the Royal Dutch Shell Technology Centre, India)
- Duties included:
- In Collaboration with the Royal Dutch Shell Technology Centre, India : Designed a feature extraction and segmentation-based algorithm to correct the inherent rotational and translationalmisalignments caused while recording Transmission electron microscopy (TEM) 2D tilt series projections.Backprojected and reconstructed the company’s corresponding catalyst crystal volumes accurately
- Worked on implementing current models and developing a faster and accurate deep network-based algorithm tostudy and improve the compressed sensing based reconstruction of Magnetic Resonance Imaging (MRI) data
- Supervisor: Professor Phaneendra Yalavarthy, Prof. Chittaranjan Hota
Projects
- End to end learning of steering commands for self driving: Nov 2021 – Dec 2021
- Description:
- Simulated self-driving car data (around 28k images from cameras mounted on the vehicle) using Udacity’s simulator along with 2 students
- Predicted steering angle and throttle based on these images using CNN and multi-task learning
- Tech Stack: Python, PyTorch
- Code and Report
- Personality Detection of Movie Characters based on Movie Scripts: Nov 2021 – Dec 2021
- Description:
- Trained and compared the performance of various machine learning models like Decision Trees, SVM, XGBoost, Multi-layer perceptron and word embeddings like BERT, Glove, Word2Vec to predict the Big Five personality traits based on social media data
- Applied the best model (Multi-layer perceptron with BERT embeddings) on movie scripts to understand the character’s emotions in each scene
- Tech Stack: Python, PyTorch
- Report
- PennCloud - A distributed platform for webmail and storage service: Nov 2021 – Dec 2021
- Description:
- Built a full-stack distributed cloud platform with webmail and storage service, analogous to Gmail and Google Drive
- Designed a highly scalable, fault-tolerant key-value data store at the backend similar to Google’s Bigtable that supports efficient consistency, replication, robust checkpointing, and recovery
- Implemented load-balancing of multiple frontend servers that users can access from their browsers
- Tech Stack: C/C++, HTML, CSS
- Report
- Input Movie Plot based Anime Poster Suggestion: Sept 2020
- Description:
- Scraped Movie and Anime titles, plots and posters data from IMDB and MyAnimeList websites
- Evaluated the performance of various Image Translation models to generate an anime version of the movie poster corresponding to the semantically similar movie plot (using Universal Sentence Encoder)
- Tech Stack: TensorFlow Hub, Scrapy, PyTorch
- Code
- Social Distancing Detector: Aug 2020
- Description:
- Detected people in a frame by finetuning Mask-RCNN model and predicted the ones who are at risky distance
- Tech Stack: PyTorch, OpenCV
- Code
- ClassPulse Web Application: Jan 2019 – Apr 2019
- Description:
- Tech Stack: React, MongoDB, Express, NodeJS, Heroku
- Machine Learning and Information Retrieval projects: Aug 2018 – Nov 2018
- Description:
- Implemented Fischer Discriminant, Perceptron classification models, and CNN for image classification [code]
- Built a movie-recommender system by implementing Collaborative Filtering, SVD and CUR approaches [code]
- Implemented IBM models and EM algorithm for Machine Translation [code]
- Built a search system which gives the corresponding recipe for the food item in the input image using Google’s cloud vision API. Implemented Tf-Idf model for ranking the retrieved recipes
- Tech Stack: Python
- Hospital Database Management System: Jan 2018 – Apr 2018
- Description:
- Lead the team project, normalized the hospital database structure and wrote SQL queries for complex retrievals
- Built a website to check the availability of hospital rooms and staff to match them with the incoming patients
- Tech Stack: Java, MySQL
- Quiz Generator: Aug 2017 – Nov 2017
- Description:
- Developed a tool that will enable one to randomly generate a quiz containing the specified number of questions from a question bank.
- Users can also edit the question bank by inserting/deleting/modifying questions as they see fit
- Tech Stack: Java
Skills
- Programming Languages: Python, MATLAB, C/C++, Java, R, HTML/CSS
- Deep-Learning Frameworks: PyTorch, Keras
- Tools: Image Processing Toolbox, OpenCV, Scrapy, SQL, MS Office, LaTeX, GitHub
Publications
Research & Teaching
Certificates & Achievements
- 5th Summer School on AI with focus on CV and ML: IIIT Hyderabad [Aug 2021]
- Introduction to Machine Learning in Production: Deeplearning.ai, Coursera [Jun 2021]
- Generative Adversarial Networks specialization (3 courses): Deeplearning.ai, Coursera [Jun 2020]
- Winner - IEEE Xplore Challenge for Researchers in India: [Nov 2020]
- Solutions architect - Amazon Web Services Course: Linux Academy [Jun 2018]
- Microsoft Technology Associate in Azure Cloud: Verzeo Edutech [Apr 2018]
- R programming training: Innodatatics USA [Aug 2017]
