Research & Projects

Designed and built Applyr, an agentic AI platform that streamlines job applications through a swipe-based interface. Leverages LLMs to autofill application forms and generate personalised, editable cover letters in real time, reducing application time by an estimated 70%. Built for mobile-first use with integrated application tracking.

Applied Trevor Nelson's framework for technology ethics to evaluate accountability, transparency, and societal impact of institutional decisions.

Architected and deployed an LLM-powered chatbot for dynamic retrieval and summarisation of research and financial datasets for healthcare-focused clients at Nephron Research.

Studied human interaction with ChatGPT | nature and frequency of questions, user perception of responses, and utility and satisfaction derived across domains of use.

Evaluated ethics in human computer interaction with a focus on dark patterns, ie, deceptive UI/UX techniques designed to manipulate user behaviour against their own interests.

Analysed how voter uncertainty affects election outcomes from a computational social choice perspective, examining strategic voting and preference aggregation mechanisms.

Built a multi-source time series ML system for real time weather scenario modelling supporting bias-adjusted forecasting for clean energy infrastructure investments at Leyden Analytics.

Conducted a quantitative analysis of market state dynamics in S&P 500 returns using a return classification framework. Daily returns were discretised into four categories (large win, small win, small loss, large loss), generating sequences of short-term market states. Analysed their statistical properties across 32 years of data, evaluating whether behavioural responses to extreme gains or losses produce repeatable reversal or continuation patterns.

Statistical analysis of COVID-19 case, death, and test time series in India. Examined temperature and humidity effects, applied exponential smoothing for forecasting, and built compartmental models for lockdown scenarios and policy implications.

Analysed platform design strategies beyond pricing for monopoly and competitive markets, examining single-homing and multi-homing dynamics and their effects on platform participation and welfare.

Replication exercise for Stock and Watson's study on changes in international business cycle dynamics, published in the Journal of the European Economic Association (2005).

Studied mechanism design in spectrum auctions, examining auction formats, bidder strategy, and efficiency outcomes in regulatory and market contexts.

Evaluated hybrid vehicle efficiency under Indian driving conditions, modelling energy consumption and performance across representative urban and highway driving cycles.

B.Tech thesis | Applied operations research methods to analyse the design and optimisation of Mobility as a Service systems, examining routing, fleet sizing, and service integration.

Designed and implemented an ML-based vehicle detection and tracking system, applying computer vision techniques to real-world traffic data.

Design, prototyping, and simulation of power modular layout for electric vehicles at the High Voltage and Power Electronics Lab, Ohio State University.