Anshuman Kumar

I am currently working as a postdoctoral associate at University of California, where my work involves applying machine learning approaches to advance scientific research. I have more than 5 years of experience working with varied data sets and using advanced analytics for data insights. Besides my expertise in computational sciences from my PhD, I have skills in Deep Learning, PyTorch, TensorFlow, Natural Language Processing, and utilizing cross-language R/Python environments for data exploration, visualization, and predictive analysis through machine learning.

I am looking to transition into roles encompassing data science, data analysis, and machine learning engineering. I bring a wealth of knowledge and a passion for continuous skill development and learning.

Postdoctoral Associate - University of California
> Development of Regression (LASSO and Ridge) and generative AI (GAN, Diffusion) models to predict the band gap of 1-D and 2-D polymers based on their composition and crystal structure.

> Enabling Clustering Algorithm (PCA, t-SNE, DBSCAN) for the screening of multi-principal elements and high-entropy alloys, which are novel materials with high performance and stability.

> Optimization and development of machine learning deep potential in Density Functional Theory (DFT) and molecular dynamics calculations.

Research Assistant - University of California
> Developed a customized docker image incorporating GPU-enabled density functional tight binding, with Plumed and Magma, on the Azure cloud platform. Utilized this image to perform molecular dynamics simulations for accurate evaluation of free energy surfaces of drug molecules.

> Analyzed the linear polarizability and second hyperpolarizability using CCSD(T) and range-separated functionals on streptocyanines using Gaussian.

> Applied electronic structure calculations in conjunction with Boltzmann transport calculations to determine the conductivity, mobility, and band structure of DNA.

> Conducted extensive evaluations of oligopeptides’ UV circular dichroism and absorption spectra by employing the time-dependent density functional theory approach using Gaussian.

> Developed and implemented a cutting-edge deep learning model to accurately predict time-dependent control fields, enabling precise manipulation of electronic transitions in quantum systems.

> Utilized the Non-equilibrium Green’s Function approach to model the ab-initio transport properties of a doped carbon nanotube network in a two-dimensional configuration.

> Experience using high-performance computing clusters (Example: San Diego access supercomputers, Texas stampede clusters, and John Hopkins’s rockfish).

Accenture
> Led a 4-person team for the development and deployment of a web application for Cisco (San Jose, CA) using Java, contributing to the successful delivery of a critical project.

> Derived insights into user behavior using Data Analysis with Python and SQL.

> Designed and implemented phishing security vulnerabilities on web browsers using Javascript.

> Translated business requirements into technical specifications.

PhD, Computational Science, University of California - 2023
MS, Electrical Engineering, University of California - 2017