Master s in Computer Science

Akshata Sawhney

858-***-**** admsv0@r.postjobfree.com https://www.linkedin.com/in/akshata-sawhney Dallas, TX EDUCATION

The University of Texas at Dallas, Texas, USA August 2019 - May 2021 Master’s in Computer Science CGPA: 3.5/4

Manipal University Jaipur Rajasthan, India August 2015 - May 2019 B.Tech. in Computer Science CGPA: 8.54/10

RELEVANT COURSEWORK

Data Structures, Design and Analysis of Computer Algorithms, Database Design, Machine Learning, Computer Vision, Data Science, Big Data Analytics and Information Retrieval. PROFESSIONAL EXPERIENCE

Computer Vision Intern, Wobot Intelligence January 2019 - May 2019

• Developed a deep learning image classification model to distinguish healthy leaves from diseased ones for several plants.

• Keras and Tensorflow were used for the detections of diseased parts of the leaves, and Flask to create a web application for internal deployment.

• An accuracy of 93.8% was achieved.

Software Development Intern, Wipro June 2018 - July 2018

• Worked on the flow model for Crime and Criminal Tracking Network Systems, an application used by the Delhi Police.

• Worked with the product management team to understand the stakeholder requirements and implemented solution in Java. Wrote unit tests and integration tests for the project. PROJECTS

Twitter Sentiment Analysis using Spark Streaming and Kafka October 2020 - November 2020

• The project performed sentiment analysis on relevant tweets in real-time based on hashtags.

• Spark streaming and Kafka framework were used to stream the Twitter data. The results were visualized using ElasticSearch and Kibana.

Multi-Label Classification of Toxic Comments using BERT April 2020 - May 2020

• The project classified toxic comments from Wikipedia’s talk page edits into categories based on their severity.

• Pre-trained BERT-base, uncased model was used and fine-tuned along with Tensorflow to build a multi-label text classifier. Achieved a precision score of 0.96.

Identification of Social Distancing violations using Deep Learning April 2020

• The project identified instances where the distance between two people was less than the safe distance.

• Transfer-learning on the Faster Region-based CNN Resnet 101 was used to develop the pedestrian detection model. An accuracy of 96.2% was achieved.

Fraud Detection and Prevention using Machine Learning September 2019 - October 2019

• The project classified banking transactions into fraudulent or non-fraudulent by analyzing buying pattern, geographical data and other metrics.

• Classification algorithms used – Logistic Regression, Decision Tree, Categorical Boosting and XGBoost. A precision score of 0.63 was obtained. Explored other alternatives to handle the issue of imbalanced data and achieved a precision score of 0.97.

TECHNICAL SKILLS

• Programming Languages: Python, PySpark, Java, C, C++

• Database: MySQL

• Techniques: Regression, Classification, KNN, SVM, Artificial Neural Networks, Convolutional Neural Networks, K- means Clustering, BERT(NLP), LSTM, MapReduce

• Libraries: Tensorflow, Keras, OpenCV, Flask, Scikit-learn, Pandas, Numpy, Matplotlib, ElasticSearch

• Tools: Hadoop, Spark, Jupyter Notebook, Visual Studio, Spyder, Git, Kibana, Android Studio ACTIVITIES

• Teaching Assistant for the CS1337 Computer Science-I course, Spring 2020.

• Teaching Assistant for the CS1336 Programming Fundamentals course, Fall 2019.

• Instructor for the UTD Computer Science Outreach Program.

Contact this candidate