Member of the Photon team

1. During my Software Engineering internship at Snowflake, I successfully implemented a Correlation Matrix function for SnowPark DataFrames. This function is capable of processing large DataFrames, which are not supported by traditional libraries like NumPy and Pandas due to RAM limitations.2. I explored and evaluated multiple techniques, including MultiThreading and Stored Procedure approach, on various sized DataFrames to optimize the performance of the Correlation Matrix function. The implementation resulted in a 4x improvement compared to Pandas' built-in correlation function.3. For my second project, I developed the Snowflake GridSearchCV library. This library leverages the power of multithreaded Stored Procedure calls to significantly enhance the performance of finding the best hyperparameter combinations for a model. On average, the library provided a 6x performance boost over the traditional Scikit-Learn GridSearchCV.

As a Grader for the Advanced Algorithms course (COMPSCI 611) under the guidance of Professors Ramesh Sitharaman and Hung Lee, I evaluated student performance by grading assignments and exam submissions. This role allowed me to gain a deeper understanding of the course material while contributing to the success of the students.

I worked as an annotator for a research project in the Information Extraction and Synthesis Lab.

1. During my Fall 2021 internship at Snowflake as a Software Engineering Intern, I was tasked with a project aimed at enhancing the functionality of Python UDF handlers. I enabled the handlers to receive inputs in the form of vectorized DataFrames and return outputs as Pandas arrays or Series.2. Further, I created a "@vectorized" decorator that allowed for the specification of input format and target batch size. Additionally, I implemented timeouts for vectorized input/output to ensure efficient performance.3. The outcome of my efforts was a marked improvement in latency, with a decrease of 3x compared to the previous row-at-a-time invocations.

1. Expressive Decision Tables (EDT) are an easy-to-use notation to specify requirements from an industrial point of view.2. I automated testing of given EDTs with specific configurations and parameters.3. I also automated the optimization testing and coverage testing of all the algorithms for a given set of Expressive Decision Tables.