Oct 2020 – Feb 2022 · 1 yr 4 mos · Mumbai, Maharashtra, India

• Developed and implemented market-making strategies leveraging multi-level order book logic to enhance trading efficiency.
• Trained recurrent neural network (RNN) models to identify directional alpha in index options, improving predictive accuracy.
• Engineered robust live production code for signal generation and execution, optimizing trade performance.
• Conducted thorough analysis of post-trade fill toxicity and execution quality, leading to improved trading strategies.

May 2019 – Jul 2019 · 2 mos · United States

• Proposed novel algorithms for joint decision making among multiple agents using reinforcement learning, and analyzed the performance of the proposed algorithms.
• Considered the case where each agent aims to optimize a joint objective function of the long-term rewards of all the agents which can be non-linear or non-smooth.
• Provided convergence proofs to the optimal function without the knowledge of the transition probability.
• To the best of our knowledge no such formal proof exists in Multi-Agent Reinforcement Learning(MARL).
• Proposed a novel Algorithm for achieving Mean-Field Equilibrium(MFE) in MARL using Posterior Sampling. Showed that the limiting distribution constitute a MFE.

May 2018 – Jul 2018 · 2 mos · Kanpur Area, India

• Incidence of Multicollinearity and Possible Violation of Weak Exogeneity Principle - Some Corrective
• measures and Application in a Recent Study (Research Project).
• Look into the reasons that led to the positive relationship between unemployment and GDP
• per capita growth as opposed to the desired. The possible causes we probed include MultiCollinearity among the covariates and violation of the weak exogeneity assumption.