I am a Data Science Intern (Auto ML) at Ascendo AI. I am involved in building and automating 500+ ML models for Flight Data Recorder in production environment using Random Forest, a predictive modeling technique. I have performed feature engineering to yield high predictive accuracy of 98%. Also worked with large time-series datasets to deliver product that saved 16 man hours per week for the team. I have experience in analyzing data and developing models in high quality Python code using ML frameworks and libraries like Tensorflow, scikit-learn, pandas & numPy.
Recently, I was a Graduate Computer Science student at University of Southern California. I have worked as a Research Assistant with Dr. Mahta Moghaddam on SoilSCAPE, a NASA Jet Propulsion Laboratory project. I was responsible for evaluating accuracy of soil moisture using Big Data Management, visualization, analysis and Data mining techniques.
Interests: Inclined to work towards data science, NLP and ML applications in the sectors of social media, healthcare and business intelligence (recommendation systems). I also see myself potentially involved in roles encompassing Machine Learning, Big Data analytics (cloud infrastructure) along with software development.
I have previously worked as a Senior Software Engineer for 2+ years. I have hands on experience in SQL, Amazon Web Services (AWS), Linux, Oracle Database and Shell Scripting, working in an Agile environment.
-
Experience
I have built and automated 500+ models (Auto ML) for Flight Data Recorder in production environment applying Random Forest, a predictive modeling technique. I have performed feature engineering to yield high predictive accuracy of 98%. Also assessed and worked with large time-series dataset to deliver product saving 16 man hours per week for team. Moreover, collaborated with product team to develop features around user behavior and making APIs employing Flask, MySQL, HTML, CSS and Javascript. Additionally, analyzed data and devised models in Python using ML frameworks and libraries such as Tensorflow, keras, scikit-learn, pandas and numPy. I eventually communicated insights through visualizations leveraging matplotlib and Tableau and deployed containerized application using Docker
-
Projects
1)Built recommendation Systems using PySpark, Apache Spark,RDD and Map Reduce. Designed Model Based, User based and Item based collaborative filtering recommendation systems. Improved speed and accuracy to 40 seconds and 97% respectively utilizing Map Reduce
2)Created a news website with functionalities namely autosuggest search, bookmarking items (addition and deletion) and sharing on social media platforms. Added a comment box for posting comments on detailed article pages along with implementing functionalities in particular toast and tooltip on icons. Used ReactJS, React Bootstrap, Node.js and AWS(S3 and EC2). This improved user experience and users accessibility to various categories of news in a location
3)Keyword Extraction from articles using Python, numPy and pandas. Applied NLP to extract most frequent occurring keywords from a dataset containing 3800 abstracts.
4)Implemented Linear Regression to predict quality of wine using Python and numPy. Performed predictive modeling to determine quality of wine by training model on various features. Accomplished binary classification using perceptron and logistic loss and multiclass classification on given dataset