Experience

Systematically identified and cleared efficiency bottlenecks via detailed benchmarking and testing for the mcmcse R package. mcmcse is the leading package for estimating for estimating Monte Carlo standard errors and reliable calculation of effective sample size. I focused on improving the critical subroutines, improving runtimes by a factor of 60, as well as migrating the codebase to Rcpp, parallel implementation of functions and improving numerical stability.

Desgined proprietary trading algorithms for futures and options, with a special focus on improving results on small holding periods. Combined technical indicators across different timeframes to improve the sharpe and return of strategies. Conducted a comprehensive study of return on small holding periods for candlestick patterns and their respones to small variations.

Involved in discussions and presentations of recent advances in probabilistic machine learning (PML) and inference under prof. Piyush Rai.

Conducted talks (by students and faculties) on popular areas of mathematics and statistics. Mentored freshers for a project on reinforcement learning. Assisted in the organization of online math treasure hunt and math murder mystery.

Working with Prof. Dootika Vats on Markov chain Monte Carlo (MCMC), with a special focus on output analysis of multivariate parallel Markov chains. Also starting work on theoretical properties of geometrically ergodic Markov chains.

Organized programming camps, hackathons and lecture series on various topics of programming, ranging from graph theory to blockchains. Mentored more than 150 students for projects on NLP models, optimization techniques and cryptoML.

Implemented SOTA online recommender engine based on collaborative filtering. Desgined a novel autoencoder based online recommendation engine. Implemented a bidirectional LSTM model with ELMO to flag hate speech in comments.