About
Ph.D. Candidate at Columbia University
&
AI Resident at Google X
I am a last year Ph.D. candidate in Urban group at Columbia University coadvised by Alexander Urban (Department of Chemical Engineering) and Simon Billinge (Department of Applied Physics and Applied Mathematics), working on materials representation learning and discovery through generative models and DFT/MD simulations.
I am also at Google X (the moonshot factory) as an AI Resident working on Computer Vision and Generative models for Molecular and Materials Discovery.
I received my B.S. from the University of California, Berkeley in 2019, where I performed undergraduate research with Mark Asta, working on defect thermodynamics with Density Functional Theory (DFT).
Besides work, you can usually find me playing chess, travelling, learning my fourth language (Can you guess the three languages that I know?) and spending time with people I love.
Resume
Below is a preview of my resume; access my full CV here. [updated Jun 2023]
Education
Columbia University
Ph.D. Materials Science and Engineering
Department of Applied Physics and Applied Mathematics
2019 - 2023 [Expected] | New York, NY
- Supervised by Alexander Urban and Simon Billinge.
- Research on ML and quantum-guided inverse design of materials with target properties.
- Coursework in Machine Learning, Computational Mathematics, Numerical Methods, Group Theory, Atomistic Simulations, DFT, Phonons.
University of California, Berkeley
B.S. Materials Science and Engineering
B.S. Chemical Engineering
Minor in Eletrical Engineering and Computer Sciences
2015 - 2019 | Berkeley, CA
- Undergraduate research in Mark Asta's group on Thermodynamics of charged defects with DFT.
- Coursework in Data Science, Circuit Design, Optics, Control Theory, Materials Science, Chemical Engineering.
Industry Experience
Google X
AI Resident
June 2022 - Present | Mountain View, CA
- Generative Models for Materials Discovery and Computer Vision
- X is Alphabet’s moonshot factory. I am part of a confidential team, working in the area of ML for materials recycling.
- Co‑developed a patent and a workshop paper (see below) for the applications related to computer vision and generative models for materials discovery.
Google X
AI Resident
May 2021 - Dec 2021 | Mountain View, CA
- Generative Models for Molecular Discovery
- Co‑developed patents (filed Oct. 2022, see below) for the applications related to generative models for molecular discovery.
The Quant Edge
Quantitative Research Intern
Aug 2020 - Feb 2021 | New York, NY
- ML for Sports Betting
- Monte Carlo simulations and statistical methods for soccer match prediction, and a ranking algorithm for horse racing predictions.
Papers
Gharakhanyan, V., Aalto, M. S., Alsoulah, A., Artrith N., and Urban, A. (2023). Constructing and compressing global moment descriptors from local atomic environments. (2023). The 11th International Conference on Learning Representations (ICLR 2023), ML4Materials workshop. openreview.net
Garrido Torres, J. A., Gharakhanyan, V., Artrith, N., Eegholm, T. H., and Urban, A. (2021). Augmenting zero-Kelvin quantum mechanics with machine learning for the prediction of chemical reactions at high temperatures. Nature communications, 12(1), 1-9. doi.org/10.1038/s41467-021-27154-2
Gadhiya, T., Shah, F., Vyas, N., Gharakhanyan, V., Yang, J. H., and Holiday, A. (2022). Directional Variational Transformers for continuous molecular embedding. ELLIS 2022 ML4Molecules workshop. ml.jku.at
Gharakhanyan, V., Chatterjee, S., Trinkle, D. R., Artrith N., and Urban, A. (2023). Constructing a compressed space of global representations from local atomic environments with information theory and autoencoders. [in preparation]
Gharakhanyan, V., Wang, T., Ramesh, S., Chatterjee, S., Trinkle, D. R., and Urban, A. (2023). ML‑accelerated molecular dynamics simulations for predicting equilibrium melting points from short non‑equilibrium simulations. [in preparation]
Gharakhanyan, V., Wirth, L., Garrido Torres, J. A., Eisenberg, E., and Urban, A. (2023). Discovering melting temperature prediction models of inorganic solids by combining supervised and unsupervised learning approaches. [in preparation]
Gharakhanyan, V., Aalto, M. S., and Urban, A. (2023). Quantifying the transferability of materials representations. [in preparation]
Wirth, L., Gharakhanyan, V., Thompson, M., Lu, Z., Wang, T., Gonzalez, D., Chatterjee, S., Urban, A. and Trinkle, D. R. (2023). Representation of free energy surfaces of binary alloy systems from CALPHAD through symbolic learning studies. [in preparation]
Honors and Awards
GDS IMPACT Award for Excellence in Graduate Research
FIP Distinguished Student Award
DMP Ovshinsky Student Prize
GERA Energy Workshop Award
American Physical Society [Mar 2023]
NSF Conference Fellowship
Mechanistic ML and Digital Twins (MMLDT-CSET) 2021 Conference [July 2021]
Data Science/Medical Research Program Fellowship
TechFoundation & Harvard Medical School [July - August 2020]
Paper on Mathematical Modelling of Viruses, President’s Special Award
President of the Republic of Armenia, Armen Sarkissian [June 2020]
Design Competition: The Energy Transition Challenge, 2nd award
Chevron Corporation, Berkeley, CA [May 2018]
President’s Annual Award for the Best Student in Information Technology
Synopsys, Yerevan, Armenia [October 2013]
Two Bronze medals, International Chemistry Olympiads 2012 and 2013
Washington D.C. and Moscow, Russia [June 2012 and June 2013]
Professional Service
Research Mentor ‑ 2 Masters, 2 Undergraduate and 2 High‑school students
Reviewer ‑ AI4Mat workshop @ NeurIPS, Journal of Chemical Physics
Patents
Gharakhanyan, V., Yang, J. H., Gadhiya, T., and Holiday, A. (X Development LLC, 2023). Search for candidate molecules using quantum or thermodynamical simulations and autoencoder. U.S. Patent Application 17/967,704, June 1, 2023.
Yang, J. H., Gharakhanyan, V., Gadhiya, T., and Holiday, A. (X Development LLC, 2023). Ionic liquid-based depolymerization optimization. U.S. Patent Application 17/967,711, June 1, 2023.
Gadhiya, T., Shah, F., Vyas, N., Yang, J. H., Gharakhanyan, V., and Holiday, A. (X Development LLC, 2023). Molecular structure transformers for property prediction. U.S. Patent Application 17/967,685, June 1, 2023.
Gadhiya, T., Shah, F., Vyas, N., Yang, J. H., Gharakhanyan, V., and Holiday, A. (X Development LLC, 2023). Molecular structure transformers for property prediction. U.S. Patent Application 17/967,685, June 1, 2023.
Holiday, A., Gharakhanyan, V., Gadhiya, T., Vyas, N., and Shah, F. (X Development LLC, 2023). Machine learning platform for finding solid catalysts for depolymerization reactions. U.S. Provisional Patent Application 63/483,807, filed Feb 8, 2023.
Holiday, A., Gharakhanyan, V., Gadhiya, T., Vyas, N., and Shah, F. (X Development LLC, 2023). Machine learning platform for generating solid catalysts for depolymerization reactions. U.S. Provisional Patent Application 63/509,220, filed June 20, 2023.
Presentations
Navigating materials design space with autoencoders to learn materials thermodynamics
APS March Meeting, March 2023, Las Vegas, NV
Combined clustering and regression for predicting melting temperatures of solids
TMS Annual Meeting, March 2022, Anaheim, CA
Teaching Experience
Teaching at Columbia University
- TA - Analysis of Engineering Problems course (CHEN 3020) [Sp2023]
- TA - Atomistic Simulations course (CHEN 4880) [Sp2021]
- TA - Computational Math: Numerical Methods course (APMA 4300) [Fa2019-Sp2020]
- Instructor - Academic Success Program - Introduction to Statistics course [Su2020, Su2021]
Teaching at UC Berkeley
- TA - Quantum Mechanics course (Chem 120A) [Fa2018-Sp2019]
- TA - General Chemistry course (Chem 1A) [Su2018]
- Head Tutor - College of Chemistry [Fa2017-Sp2019]
Teaching elsewhere
- Instructor - Workshop on Chemical Process Control and Dynamics at Tumo Center for Creative Technologies in Yerevan, Armenia [Wi2018]