Hi! I'm Shivani Soman, a Data Scientist based in New York City, NY.
Working in the finance field, I translate seemingly uninformative data into valuable and actionable insights. I build concise and insightful reporting dashboards and use machine learning solutions to improve clients' decision making power. My goal is to build innovative data-driven solutions that would help optimize everyday processes.
In my free time I enjoy impressing people with my ukulele skills, playing video games and trying to get my rubik's cube solve time under 30 secs.
Relevant Coursework : Big Data Analytics, Database Management and Statistical Computing, Bioinformatics, Large Scale Data Mining, Learning and Reasoning with Bayesian Networks, Health Analytics. GPA : 3.93
Relevant Coursework : Data Mining Techniques and Applications, Business Intelligence, Machine Learning, Data Structures and Algorithms, Database Systems , Operating Systems, Computer Networks. Final Percentage : 71% (First Class with Distinction)
Developed a two-fold method to classify music into 10 different genres using Convolutional Recurrent Neural Networks (CRNN) for spectrogram analysis and traditional machine learning classifiers for analysis of Mel-Frequency Spectral Coefficients (MFCCs) derived from the audio samples with an accuracy of 86%.
View ProjectPerformed predictive analysis on huge amounts of metagenomic data (> 3 TB) to correctly predict the origin of each metagenomic sample using neural networks and XGBoost.
Built a deep learning algorithm using Long Short-Term Memory Networks (LSTMs) by collecting data from various sensors like accelerometer and gyroscope obtained from a smart watch to accurately predict the activity the user was performing like standing, sitting, walking, etc.
View ProjectAnalyzed millions of tweets before and during Super Bowl XLIX (2015) to determine how and why public sentiments towards the New England Patriots and the Seattle Seahawks changed over the span of the championship match.
View ProjectDeveloped a Machine Learning model using NLP that classifies genetic mutations of cancer genes from an expert annotated knowledge base and text-based clinical literature into a set of predefined classes. Obtained a better score than the 1st rank on the leaderboard of this Kaggle competition using Word2Vec Embeddings and LightGBM.
View Project