Junaid Hasan
Junaid Hasan
I am a PhD Candidate at the University of Washington, Seattle, working under the supervision of Farbod Shokrieh.
My research focuses on Tropical Geometry, exploring algebraic curves through a combinatorial lens. My current work involves energy minimizers on metric graphs and the study of tropical curves.
Apart from Math I am passionate about Data and ML, Linux hacking, HomeLab Setups, Photography to name a few.
Investigating the connection between graph divisor theory (chip-firing) and resistance curvature. This work introduces a discrete Foster energy uniquely minimized by the canonical break divisor on positively curved graphs.
Establishing a graded-ring isomorphism for a tropical Dolbeault theorem. By defining cup and cap products on singular homology, we establish a formula for computing volumes on tropical abelian varieties.
Extending classical tools to work with integral coefficients. Constructing a full Beauville decomposition for the Chow ring and defining an SL2 representation over controlled denominators.
An empirical study of query optimization and execution architectures (Vectorized vs. Volcano) using a 41-million row NYC Taxi dataset. Demonstrated an 18,000x difference in ingestion latency between row-stores and zero-copy columnar engines.
A Deep Learning project for New Atlantis Labs using autoencoder architectures and SHAP analysis to identify predictors of oxygen depletion in the Gulf of Mexico. Achieved an R² of 0.9991 when combined with XGBoost.
View Certification →Developed predictive models using historical power outage records and NOAA weather data to forecast regional utility failures across the United States based on storm strength and historical insights.
View Certification →An interactive D3.js sunburst visualization of 2M+ miniature games from the Lichess database. The board is clickable, allowing users to compare their moves against top historical trends to learn common openings and traps.