I am an independent machine learning and computer vision researcher based in India, focusing on geometric shape analysis, graph matching, and computational paleography.

Currently, I am building Project Vatayan—a 5-tier computer vision pipeline (combining Zhang-Suen skeletonization, Directed Hausdorff Distance, Graph Edit Distance, Hu Invariant Moments, and Euler topological isomerism) to objectively test structural continuity between the undeciphered Indus Valley Script (641 signs) and early Iron Age graffiti from the Vaigai River / Keeladi excavations (42 deduplicated base signs).

All datasets, skeletonized assets, and code are open-source:

- Preprint & Methodology (Zenodo): https://zenodo.org/records/23859986

- Master Pipeline & Datasets (Kaggle): https://www.kaggle.com/arkapravac366