Research &
Projects.

Benchmarked Dijkstra vs. A* on a 302,066-node graph of Lucknow's road network (4,132 km²). Built a high-performance C++17 pipeline using Libosmium to parse OpenStreetMap PBF data. Identified the Riverine Bottleneck Paradox and Cantonment Regularity Effect. A* achieved a mean search-space reduction of 78.13% (p < 0.0001) across 1,000 randomised trials.

Collaborative computational study. Designed a theoretical three-channel rGO-aptamer chip for simultaneous detection of arsenic, fluoride, and lead in rural Indian groundwater. Authored the Python simulation engine (adsorption isotherms, EIS, Monte Carlo LOD propagation, Random Forest classifier). Open-source on GitHub.

C++ engine normalising Olympic sprint times (100m, 200m, 400m, Tokyo 2020 & Paris 2024) for wind, altitude, temperature, humidity, lane curvature, and track surface. Proposed Physics Adjusted Time (PAT) for World Athletics. Discovered the Jacobs-Lyles Paradox. OmniSprint API, Python library, and public website in active development.

Full-stack AI emotional-support app for adolescents navigating grief. Features a custom-prompted AI companion (Mira), grief journey tracker, breathing regulation engine, memory gallery, teacher note generator, and emotional analytics. Built on glassmorphism UI with three adaptive themes. Submitted to HackerBoundary Global AI Hackathon.

Graphical interface for one of the most powerful RSA crackers built specifically for capture-the-flag competitions. Contributed to the open-source CTF community, making advanced RSA attack vectors accessible to the community.

Web app that converts typed text into realistic handwriting. Live and publicly accessible at likhavatth.netlify.app. Built independently and deployed on Netlify.