Hello, I am Zachary Gariepy



Data Science | Business Intelligence | Analytics | Machine Learning

Technical Skills

Languages

Python, SQL, Bash, R, MATLAB, LaTeX, HTML/CSS

Tools & Platforms

Jupyter, Docker, Linux, AWS (Lambda, Glue, S3, Athena), Git

BI & Analytics

Tableau, Salesforce CRM (Analytics Studio), Google Analytics (GA4), Looker, Klipfolio, Power BI, JMP, Blender, Ovito, VESTA

Data Packages

NumPy, Pandas, Matplotlib, Seaborn, Pytest, Plotly, SciPy, Atomic Simulation Environment (ASE)

Machine Learning Packages

TensorFlow/Keras, Pytorch, Pytorch Geometric, GpyTorch, Sklearn, RDKit, NetworkX

Simulation Software

Vienne Ab initio Simulation Package (VASP), Quantum Espresso (QE), LAMMPS, COMSOL

Skills

Business process model and notation (BPMN), unified modeling language (UML), REST API's, object-oriented programming, exploratory data analysis, high-throughput computations, predictive regression/classification, numerical machine learning, dashboards, reporting business insights and analytics

Data Science & ML Projects

An Exploration of NBA Team Building Strategy Using Economic Theory & Statistics

Data wrangling from ESPN + NBA APIs, SQLite3 db compilation for expedited querying, analysis using Lorenz curves, Gini coef., Thiel indexes, Principal component analysis and more. Check out the resulting Medium Article.

Automatic Graph Representation Algorithm for Graph Neural Networks

PyTorch Geometric and NetworkX based graph generation framework for end to end conversion,training and evaluation of SQL db stored GNN datasets

Neural Network Decomposition for Empirically Derived Insights

High-throughput db generation combined with a scalable and versatile feature engineering algorithm to describe local chemical environments, train a MLP NN, decompose the model and partially reconstruct layers to gain insights into key properties

Machine Learning Powered Bottom-up Material Design

Strategic feature space design for facile optimization of material structures through supervised learning and DFT calculations.

Publications

Topic: Data Science, Machine Learning, Catalysis

Z. Gariepy et al., Automatic graph representation algorithm for heterogeneous catalysis, APL ML, 2023

Z. Gariepy et al., Machine learning assisted binary alloy catalyst design for the electroreduction of CO2 to C2 products, Energy Advances (RCS), 2023

Z.W Chen, Z. Gariepy et al., Machine-Learning-Driven High-Entropy Alloy Catalyst Discovery to Circumvent the Scaling Relation for CO2 Reduction Reaction, ACS Catalysis, 2022

Z.W Chen, Z. Gariepy et al., High-throughput and machine-learning accelerated design of high entropy alloy catalysts, Trends in Chemistry (Cell Press), 2022


Topic: Perovskites & Lithium Salts

A. Dumont, E. Nicholson, Z. Gariepy et al., Restructuring and Reshaping of CsPbX3 Perovskites by Lithium Salts, Advanced Materials Interfaces, 2022


Topic: Semiconductor & Carbonaceous Membranes

J. Patil, A. Jana, Z. Gariepy et al., Conductive carbonaceous membranes: recent progress and future opportunities, Journal of Materials Chemistry A, 2021


Education

University of Toronto, MASc.

Department: Materials Science & Engineering (2021-2023)
Thesis: Machine Learning and Data Driven Materials Design
Lab: Computational Materials Lab
Supervisor: Chandra Veer Singh
CGPA: 3.9/4.0

University of Waterloo, BASc.

Degree: Nanotechnology Engineering Honours (2016-2021)
Achievements: Dean's List, 2 years of co-op experience in startup building, quantitative analysis and microfabrication
CGPA: 3.9/4.0

Work Experience

Business Intelligence Data Analyst
GameAddik

  • Duration - 07/2023 - 02/2024
  • Reporting & automation - Conversion metrics discrepancy investigation, reporting, insights and anomoly detection automation between client and internal databases connected via Looker and REST API
  • BI & dashboarding - Developing live data Tableau dashboards for key website conversion metrics. Identifying primary conversion pipelines to guide future business and graphical design decisions
  • Data integration & analytics - Enhanced functionality of key client AWS pipelines (Glue/Lambda) by integrating additional metrics and optimizing connectivity for seamless integration with BI software solutions.

Data Scientist Graduate Co-op
IPEX Technologies

  • Duration - 04/2022 - 10/2022
  • Database unification - Automated extraction and processing of siloed formulation repositories for streamlined data queries and accelerated DOE generation
  • Polymer formulation optimization - Discovered new top performing formulations through bayesian optimization. Experimental polymers yielded 80-160% improved mechanical properties
  • High-throughput testing guidance - Guided technician test plans through principal component analysis of formulations to identify gaps in formulation combinations below a given synthesis costs

MASc. Machine Learning Thesis Researcher
University of Toronto, Computational Materials Lab

  • Duration - 05/2021 - 06/2023
  • Graph neural network representation - Developped and published a graph representation algorithm for GNNs that outperform Meta AI algorithms at a 800% reduced computational cost and 25% accuracy improvement on sample databases
  • Bayesian modeling and techno-economic optimization - Created a framework that applied synthetic minority over-sampling and gaussian regression to a DFT based Hydrogen electrode dataset generated with high-throughput computations. Random sample optimization yielded a 90% improved overpotential compared to industry benchmark materials at a 15% reduced synthesis cost
  • Empirically derived mechanistic research - Published a MLP NN decomposition and reconstruction technique to reduce model abstraction and analyze KPIs and feature space importance on datapoint prediction.This method helped design a best in field CO2 conversion catalyst

Process Engineer Co-op
MIT, Grossman Group

  • Duration - 1/2020 - 10/2020
  • Business development & product design - Scaled the fabrication process of a semiconductor membrane technology and applied to incubator program for early stage startup company (Currently SiTration). Pitched funding to investors resulting in admittance to Greentown Labs and $2.35 million in Pre-Seed Funding
  • Microfabrication process engineering - Metal assisted chemical etching for first ever steric filtration membranes via largest aspect ratio (200:1) pore etching in literature
  • Failure analysis - MD LAMMPS modeling of novel filtration technology for biofouling mechanics and failure modes

Product Design - Polymer Engineer Co-op
Apple Inc.

  • Duration - 08/2018 - 04/2019
  • Project coordination - Spearheaded prototype material development as main coordinator to international collaborators for large scale biomaterial testing and analysis efforts resulting in 25% increase in recycled plastic content of iPhone 12 antenna splits
  • Rapid prototyping - Audio product hand carry program manager for emergency road-block problem solving through failure analysis and chemical reformulation of production phase coatings
  • Project reporting - High level reporting and direction advice on multiple product lines presenting to senior management, design teams and international vendors

Clean Energy Research Assistant
University of Waterloo, Yu Group.

  • Duration - 04/2017 - 08/2017
  • Process optimization - Improved green graphene synthesis methods for improved economic and environmental feasibility through gum arabic intercalation and thermal shocking
  • Advanced materials characterization - 2D material testing with Raman, SEM, TEM, FTIR, and more
  • SOP writing - Lab SOP documentation and verification of ASTM/ISO standards for publication data

Presentations & Collaborations

NAM28 - Sponsored Speaker (2023)

  • Conference: North American Catalysis Society (NAM) bi-annual meeting
  • Presentation topic: Network decomposition technique for data driven design of high entropy alloy catalysts for the CO2RR
  • Audience: Commercial sector, academia (professors & graduate students), government agencies (national labs & institutes)

NRC - AI4D Initiative (2021-2023)

  • Aim: AI powered design of high entropy alloy catalysts for the conversion of CO2 into fuels
  • Collaborators: NRC Research Officers, University of Toronto, University of Ottawa, University of Waterloo
  • Outcome: 2 publications detailing the creation of a FeCoNiCuMo HEA catalyst with best in field CO2 conversion to CH4 kinetics

Thesis Presenter - Ursula Franklin Research Seminar (2022)

  • Presentation topic: Graph Neural Network frameworks to leverage public databases
  • Audience: UofT professors & engineering graduate students

Personal Interests

Four Pictures with Captions
Image 1
Surfing
Image 3
Snowboarding
Image 2
Basketball
Image 4
Camping

Contact me

Please feel free to connect