- Building well-tested, efficient, reusable scalable software that implements financial quantitative analytic models using Python, Java, Linux Shell scripts} for over 10,000 securities monthly including debts, derivatives, asset-backed securities etc.- Subject matter expert for Investment and Capital Markets synthetic instruments.Responsible for creating, developing code to calculate their accounting values and provide support to business users and downstream teams who use these instruments.- Implemented parallel runners to forecast multiple horizons accounting analytics} at the same time, reducing average forecasting run time by 80% (for example, from over 20 hours to 4-5 hours for 63-month forecasting runs).- Extensive usage of DevOps tools and processes like Git, Jenkins, Docker and Kubernetes to build and streamline the software development process and accomplish efficient continuous-build and continuous-deployment capacity.- Generated reports using Python scripts such as annual balance sheets of Freddie Mac and DFAST reports for FHFA.- Developed Python tools for teams to compare runs results, regression results and generate cash flow files.- Responsible for writing complex unittest cases by using Pytest and Junit, integration tests to accomplish 100% code coverage.- Participated in agile ceremonies like sprint and PI planning, daily stand-ups, agile retrospectives and maintaining tasks in JIRA.

. Created Deep neural network models (with dropout, L1 and L2 regularization) to predict power system contingencies.. Generated power system contingency data using grid2op framework and analyzed real-time power grid data.. Power Grid Contingency Analysis with Machine Learning: A Review and Prospects Sam Yang, Bjorn Vaagensmith, Deepika Patra - selected for oral presentation at ResWeek 2020.

. Worked as a Graduate Assistant in Foundation of Algorithms(CSE 551) course.· Assisted professor in grading assignments, exams and quizzes.

· Extracted AT&T network device and router data from Oracle DB using JDBC, ETL using Core Javaand Pl/SQL and loading back into database for users.· Implemented data pipeline for AT&T network device and router data for distributed data processing and generating user billing statements.· JDBC connector extracted data from multiple AT&T usage servers to the IBM processing servers for data processing.· Processed raw data using Core Java and PL/SQL routines on a daily schedule in production.

Computer Vision project : Writer Identification from hand-written musical sheets· Implemented feature extraction from handwritten musical sheets using Local Gradient Histogram routine in C++· Used Hidden Markov Model (HMM) for classifying musical sheet writers.