I am currently conducting research at the Viral Communications Lab under Professor Andrew Lippman, focusing on developing generative video simulations and predictive news forecasting. My work utilizes advanced AI techniques, including multimodal AI for video and text synthesis, as well as Retrieval-Augmented Generation (RAG) and Reinforcement Learning from Human Feedback (RLHF) to improve contextual accuracy and alignment with human preferences. Additionally, I incorporate continual learning approaches to enhance strategic planning, decision-making, and policy development in response to evolving data trends.

Was part of the Facebook Blockchain Team which on June 18th, 2019 launched a new cryptocurrency Libra and programming language Move to execute custom transactions and smart contracts on the Libra Blockchain.Improved expressibility and reduced complexity in the Move IR, the syntactic frontend for Move VM and Bytecode. Worked on Parallelizing execution of transactions in the Move VM.Parallelizing Signature Verification led to a 40% increase in throughput and improved the transaction per second (tsp) from 7000 to 11,000.

Worked on a case study for Obsidian to build a parametric health insurance platform for our clients at the World Bank. Introduced a foreign function interface (FFI) in Obsidian with Java as guest language. An FFI can be used to build an Events mechanism to post events on the blockchain.

Built an decentralized web application (dApp) on Ethereum using uPort Authentication, Infura blockchaininfrastructure, Web3 and Truffle frameworks.The app can be accessed using uPort's unique identity feature and has applications such as a shared To Do list, a tamper proof feedback form, etc.

Collaborated with the Computer Vision Team and the Checkout Team at Mashgin.Built a scalable model for anomaly detection that would raise an alert using the data from self-checkout kiosks which use computer vision technology to identify goods.Generated synthetic data which was fed to a Convolution Neural Network (CNN) to distinguish between delivery boxes in a "Room Scale Inventory Tracking" project.

Collaborated with the Global Product Strategy Team with developing predictive models for their mutual funds.Used predictive analysis to estimate the redemption rate in their product for the next two years using R.

Worked on a project that uses solar panels to power pumping of water to rural areas which have erratic or no power supply.Designed a software in C++ to calculate the required wattage of solar panels to operate the water pumps.