Jeet Mozumdar

PUBLICATIONS (PREPRINTS): (1) Mozumdar, J. (2026). Entanglement Reachability and Structural Confinement Under Local Quantum Dynamics. https://doi.org/10.5281/zenodo.21846211 (2) Mozumdar, J. (2026). Compressed Quantum Dynamics and Emergent Tensor-Network Geometry in Operator Systems. https://doi.org/10.5281/zenodo.20687402 (3) Mozumdar, J. (2026). Lieb--Robinson Bounds and Entanglement Limits in Quantum Dynamics with Finite-Dimensional Memory (1.0). https://doi.org/10.5281/zenodo.19842(970. (Full-time)

1. Architected equivariant Graph Neural Networks and Deep Learning systems to develop AI-enhanced force fields, engineering multi-GPU training and C++ deployment via LibTorch for native integration into MD simulation engines. 2. Designed dynamic Hidden Markov Model architectures for longitudinal data, modeling time-varying latent processes in MATLAB & R, and improving prediction accuracy 25% over stationary HMMs. (Full-time)

• Researched phylogenetic inference across multiple courses, using MrBayes, BEAST2 and MCMC-based posterior tree distributions; extended the Conditional Clade Method (Larget et al.) by relaxing inter-clade independence assumptions; applied computational techniques from electrical engineering to model correlated state spaces. • Developed a proposal applying recurrent event methods in survival analysis to Fetal Alcohol Syndrome RWD. • Tutored statistics at MLRC, strengthening applied modeling and ...

Excluding June 2015 - Feb 2016.

Achieved Microsoft Certified Professional (MCSD) certification specializing in Windows virtual memory architecture, kernel-level thread synchronization primitives, and binary inter-process communication (IPC) paradigms (1998).