I'm a Bachelors Student from Indian Insitute of Technology (Delhi). I have experience working in the fields of Machine Learning, AI, and development. My primary interest lies in the fundamental and technical analysis of the financial markets. I indulge in Machine Learning through various projects and am an active participant in competitive programming primarily to improve my problem solving skills and my algorithmic approach towards practical applications. I have developed strong technical and organisational skills through internships at intensive research hubs like University Of Sydney. I've worked extensively in the field of Data Science, but am always open to exploring new and exciting avenues.
Civil & Computer Science Engineering
Percentage: 93%
Worked on various data structures and problem solving paradigms like Greedy, Dynamic Programming, Bitwise operations, Sparse Table, Fenwick Trees, Segment Trees, Minimum Spanning Trees etc. Made video editorials for ongoing contest problems given by thousands of programmers. Taught important DSA concepts and their implementations.
Implemented reinforcement techniques like Q learning and policy iteration using MCMC sampling and parallel tempering on atari games like pong and cartpole. Worked on Natural policy gradient, Stochastic gradient descent, likelihood ratio policy gradient, Vanilla policy gradient, Actor-critic and proximal policy optimization using clipping objective.
Impact of Lockdown in Air Contamination Levels of New Delhi
Established and maintained an effective risk-free portfolio of Europian options by using Black Scholes model to calculate the delta every tick and rapidly rehedging on the basis of that delta, assuming no transaction cost. The underlying asset is modelled using Geometric Brownian Motion and the derivative pricing is done using binomial trees. Also wrote a live graphing class using Matplotlib to visualize the options.
Wrote a python script to run on live video streams to detect whether the person is wearing a face mask with the test data accuracy of over 96 percent. The framework was built using Keras, TensorFlow, OpenCV and MobileNET. The neural network used binary cross entropy for loss and was trained on over 4k labelled images.
Runner Up in Microsoft Codefundo++ for this project. Using the characteristic features of P-waves which are faster and less damaging than S-waves of an Earthquake, predicted which buildings would be able to withstand the Earthquake by predicting the magnitude of the S-waves using an industrial standard of safety. The project supported the Ollen and Alson theory of rupture process which states that P-waves have a correlation with the magnitude of S-waves.
Harnessed Google’s BERT model for contextual language processing on publicly available SEC 8-K forms. Predicted and correlated the SnP 500 by overcoming drawbacks of directional models with MLM and NSP.
Consulted on how Carlsberg can further disrupt traditional on-trade channels in order to digitally engage with consumers and increase revenue growth to 5 percent in Europe by 2022.
Played over 5k hands and amassed over $17000 in online and live poker tournaments.
Elected to grow and nurture the debating culture in a hostel with 500+ students. Participated and broke in internal and external tournaments across Delhi.