ISHITA KHATRI

adxo0r@r.postjobfree.com +1-858-***-**** San Diego, CA LinkedIn GitHub

EDUCATION

UNIVERSITY OF CALIFORNIA SAN DIEGO, La Jolla, CA Expected Mar 2024 Master of Science, Data Science GPA: 3.73/4.00

o Relevant Courses: Machine Learning, Statistical Models, Algorithms for Data Science, Probability & Statistics in Data Science, Numerical Linear Algebra, Scalable Data Systems, Visual Learning, Data Science in Biomedicine VELLORE INSTITUTE OF TECHNOLOGY, Vellore, India Jun 2020 Bachelor of Technology, Information Technology GPA: 8.58/10.00 o Relevant Courses: Data Structures, DBMS, Operating Systems, Big Data Analytics, Artificial Intelligence, Data Mining SKILLS

Language/Tools: Python, R, Java, MATLAB, SQL, Spark, Linux, AWS, Git, Kibana, AppDynamics Libraries/Packages: Pandas, NumPy, Scikit-learn, TensorFlow, PyTorch, Keras, Dask framework Concepts: Computer Vision, Natural Language Processing, Forecasting, Regression, Classification, Deep Learning, Data Analysis EXPERIENCE

SMARR LAB, UNIVERSITY OF CALIFORNIA, SAN DIEGO, United States Machine Learning Researcher March 2023 – Present

o Working on Women’s health project which aims to analyzes different modalities and build a classification model that determines early onset of pregnancy using Machine learning algorithms o Using Time series analysis to find pattern in dataset containing cyclic and acyclic women BARCLAYS PLC, PUNE, India

Graduate Analyst, Application Support Analyst Aug 2020 – Aug 2022 o Undertook GAP release project and migrated live applications - RSA, IDV, Appian, and Fraud Case Management to aPaaS V4 environment and provided service to existing 10 million GAP customers. o Worked on an Identification and verification project hosted on the AWS platform for US credit cards wherein different functionalities like customer authentication and verification were handled and implemented monitoring jobs for these systems. o Recovered live fraud applications to avoid customer impact and kept upgrading them through patching for better performance o Carried out Disaster Recovery planning and testing for 8 fraud applications bi-yearly. ACADEMIC PROJECTS

Anime Colorization using Conditional GANs

o Designed and implemented a computer vision model which uses concepts of conditional GAN in image-to-image translation and perform anime colorization. Utilized U-Net based architecture for generator and convolutional PatchGAN classifier for Discriminator. Alternated gradient descent step on discriminator and generator used Adam optimizer with learning rate of 0.0002 o Used Mean absolute error loss which measures the pixel-wise difference between the generated and ground truth images, and adversarial loss, which captures the difference in distribution between the generated and real images. Trained 14000 images and tested it on 3500 images.

Cryptography

o Developed and implemented a secure data encryption system using cryptographic algorithms such as ECDH, RSA, and Diffie-Hellman and conducted a comparative analysis of the encryption time, decryption time, and memory utilization of all algorithms. o Visualized the results through informative graphs and discovered that ECDH outperformed RSA and Diffie-Hellman, with an average encryption time of 1 second, decryption time of 0.0015 seconds, and 11.7 kB memory usage. Designing a Twitter Sentiment Analysis and Recommender System using Machine Learning Algorithms o Developed and proposed a sentiment analysis system that can classify tweets as positive, negative, or neutral based on a dataset of 100,000 tweets. Demonstrated expertise in machine learning algorithms such as Tf-idf, Random Forest, Naïve Bayes, Collaborative Filtering, Snowball Stemmer, and K-nearest neighbor by applying them to both Kaggle and Twitter API datasets. o Conducted an accuracy comparison of the various algorithms and recommended the users to follow or block based on polarity using a threshold of 0.5

Identification of Fake News using Machine Learning Techniques o Developed and designed a semi-supervised system for identifying fake news using cutting-edge technologies and techniques by pre- processing a large dataset of 250,000 news articles sourced from Kaggle to prepare the data for training and testing. o Generated a self-learning system using the LSTM model to improve the system's accuracy over time. Utilized SVM and Naive Bayes algorithms for classification of news articles and achieved a highly accurate system with a precision rate of 93%. College Predictor System

o Designed and implemented a predictive system for assessing the likelihood of admission to a university based on historical records and using advanced techniques such as K-means clustering and Silhouette Score. o Utilized Backpropagation with gradient descent to determine the weights of the system and accurately calculate the probability of acceptance to specific universities. Applied multilayer perceptron algorithms to predict the tier of universities and generated a comprehensive list of recommended colleges based on a dataset with 15,000 entries. PUBLICATIONS AND ONGOING WORK

o Designing A Twitter Sentiment Analysis and Recommender System Using Machine Learning Techniques link o Identification of Fake News Using Machine Learning Techniques link