Responsible for conceptualizing and implementing market-neutral, medium-frequency quantitative strategies resulting in excess returns (alpha). Surveying financial literature and analyzing data from various sources to seek out sources of alpha following from inefficiencies and persistent causal relations in market behavior, and, aggregating these signals into robust models, with the objective of achieving high returns and low turnover alphas.

Assessment of risk culture in financial institutions using news media:Worked in cognitive computing division to develop an automated method for assessing risk culture in financial institutions by analyzing news articles. Devised and implemented a pipeline, trained using machine learning algorithms, which processes news corpus and comes up with risk culture scores for given financial institutions. These predicted scores for test data-sets matched the ground truth with high correlation. My work has been accepted for being published in 16th Finance, Risk and Accounting Perspective(FRAP) Conference hosted by Oxford Research Academy.

Rewrote the existing Java single machine code for GBDT(Gradient Boosting Decision Tree) algorithm to CUDA for execution on parallel architecture(NVIDIA GPU) to reduce the overall training time. Experimented with and applied various optimizations to reduce the training time by 94% (100 secs to 6 secs per tree). Offered full time offer based on exceptional performance during internship.