I built the new stats engine and the enhanced reporting system at VWO with over 15 new features for cutting edge experimentation.A summary of my work is here: https://vwo.com/why-us/technology/statistics/

Leading the Data Science Team to build a new statistically advanced reporting system for VWO. Some of the advancements we have made to VWO's core statistics are as follows:- My team optimized the calculation speed of the legacy Bayesian system at VWO by over 10,000x (two improvements of 100x compounded).- My team has increased the accuracy of the internal statistics exponentially by solving the peeking and multiple comparison problems in experimentation.- We conceptualized, developed, and productized over 15 new statistical features to VWO reports after talking to multiple customers and influencers in experimentation.

◦ Bayesian Vehicle Crash Risk: Developed a Bayesian Model to predict crash risks for all road segments of Washington, DC using pymc3.◦ Bayesian COVID-19 Analysis: Trained a Bayesian Model to answer quantitative questions about the emergence of COVID-19 in various US States.

- Applied Reinforcement Learning on Online Courseware Data.- Built a Regression Approach for identifying ineffective components of coursework.- Conceptualized a methodology for performance-based personalized content recommendation.

◦ Dynamic Personalisation: Developed an algorithm for dynamically personalised conversation rate optimisation.◦ Generalised Bayesian A/B Test: Conceptualised and implemented Generalised Bayesian A/B Tests for various conversion metric optimisations.◦ Monte Carlo Integration Optimisation: Statistically optimized Bayesian A/B test calculations by a factor of 100 sharply reducing server costs for the company.

◦ Random Forest Based Stop Losses: Trained Random Forest Based Stop Loss predictors increasing stock trading returns by 30%.

I thoroughly studied the field of associational analysis and in the process learned about the limitations of the Pearson Correlation Coefficient. In an attempt to overcome those limitations, I developed a novel approach to identify associations among variables using Bayesian Statistics. The developed approach outperformed the existing approaches by almost two folds and was published in PLOS One.

My responsibilities as a software engineer in InfoEdge include.1. Coding and development of services in Java and Php.2. Managing data in MySQL and finding ways to make it more organized.3. Learning new concepts and technologies and sharing it with employees across my team and the company.

As an intern in Oracle, I worked on a 6 month project targeted at automating the process of testing Oracle Retail Suite. We used different tools and languages including Synergy(an in-house tool developed by Oracle).