cv
List of related skills, education, and experiences.
Updated 28/12/2023.
Education
-
2020- PhD in Theoretical Physical Chemistry
University of Toronto, Toronto, ON Department of Chemistry - Affiliated with the Vector Institute for Artificial Intelligence.
- Supervisor: Alán Aspuru-Guzik.
-
2017-2020 MSc in Physics
University of British Columbia, Vancouver, BC Department of Physics and Astronomy - Supervisor: Sarah A. Burke.
-
2013-2017 BSc (Hons.) in Physics
McGill University, Montréal, QC Department of Physics - Graduated first-class honours.
- Minor in Interdiscplinary Life Sciences
- Supervisor: Peter Grütter.
Experience
-
5/2020- PhD graduate researcher
University of Toronto, Toronto, ON Matter Lab - Creating a software package, DIONYSUS for probabilistic prediction and classification on small datasets of molecules using TensorFlow. Examining the calibration of uncertainty in deep learning models. Results published in RSC Digital Discovery.
- Compiled kernal-based prediction and classification ML package for chemistry called GAUCHE. Co-first author of results published in NeurIPS 2023.
- Implementing string-based generative models in PyTorch for a molecular design benchmarking project, TARTARUS. Results published in NeurIPS 2023.
- Developing variational autoencoders trained with Bayesian models for low-data inverse design of molecules.
- Designing new molecular graph features based on electronic structures with the goal of improving performance of graph-based neural networks on chemical structures.
-
5-8/2023 AI PhD Resident
SandboxAQ, Toronto (Remote), ON - Joined the drug development and chemical simulations team (AQBioSim). Worked on both internal research and client-facing projects.
- Performed active learning guided high-throughput virtual screening of ~30,000 small organic compounds. Our virtual screening platform found almost 1000 possible drug candidates for our client's target protein.
- Developed a 3D equivariant probabilistic graph neural network (GNN) and an efficient protein-ligand complex extraction algorithm for prediction of free energy perturbation calculations (FEP) binding affinities. Further demonstrated the model's effectiveness in active learning experiments for drug hit identification.
- Work presented at the NeurIPS 2023 AI4D3 workshop, with a parallel submission planned for a peer reviewed chemistry journal. Work is part of a pending patent.
-
2017-2020 MSc graduate researcher
University of British Columbia, Vancouver, BC Laboratory for Atomic Imaging (LAIR) - Studied the luminescence of single organic photovoltaic molecules using scanning tunneling microscopy (STM). Presented findings as an invited speaker at IVC 2019.
- Supervised and provided guidance to a summer student in classifying molecules in STM images using computer vision and neural networks.
- Performed DFT (Gaussian16) and quantum chemistry calculations on small molecules for comparison with experimental results.
- Characterized novel organic deep-blue fluorescent molecules designed by the Hudson Group at UBC.
- Developed MATLAB analysis script to study strain in graphene samples, characterized by STM. Results published in Science Advances.
Skills
-
Programming
- Proficient in Python, MATLAB, IDL, and LabView. Experience with FORTRAN, Java, and C.
- Familiar with Linux environment and the command line. Basic experience in high-performance computing (SLURM).
- Experience with using Git version control. Strong LaTeX skills.
-
Machine learning and data science
- Data wrangling, analysis, and visualization in MATLAB, IDL, and Python (numpy, matplotlib, pandas, and seaborn).
- Highly experienced with common ML frameworks in Python: TensorFlow/Keras, PyTorch/PyTorch Lightning, and scikit-learn.
- Experience with probabilistic frameworks such as GPyTorch, and TensorFlow-Probability.
Other Interests
- Musical theatre - watching and performing.
- Violinist in the Strings Attached Orchestra.
- Hiking, skiing, and snowboarding.