Tim Baker Profile Pic

About Me

As an aspiring AI developer, I'm passionate about using machine learning tools to innovate the world at large. This past co-op term, I interned at Nuralogix, where I developed machine learning models to predict blood pressure from features derived via transdermal optical imaging (TOI). Prior to that, I worked at Environment and Climate Change Canada (ECCC), where I used deep learning to predict precipitation type from time series data. I also worked at PerkinElmer Inc, where I helped develop an AI solution for drug discovery. In the summer of 2018, I interned with Sunnybrook's Focused Ultrasound group to help develop treatments for cancer and dementia patients. The following year, I became a summer student at the Schulich Heart Centre, where I co-authored a paper on using MRI to diagnosing heart arrhythmias.

Contact Details

John Quinto
University of Waterloo
Waterloo Ontario,
289-654-4930
jquinto64@gmail.com

Download Resume

Education

University of Waterloo

Candidate for BASc in Biomedical Engineering •Class of 2023

Relevant Courses: Data Structures and Algorithms; Digital Computation; Statistics and Experimental Design; Linear Signals and Systems; Digital Systems; Biomedical Engineering Design; Prototyping, Simulation and Design; Circuits, Instrumentation, and Measurements; Human Factors in Biomedical Design

Work

Nuralogix

Data Science Software Developer• May 2021 - Present

At Nuralogix, I trained a transformer-based neural network to predict blood pressure from facial biomarkers, which achieved 5% lower root-mean squared error (RMSE) than previous regression models. I also implemented a feature selection method that cut computation times from 6 hours to 10 minutes, while reducing the error of an XGBoost ensemble by 6%. Moreover, I developed a CNN-based facial recognition and face clustering pipeline that identified unique participants in the training and testing datasets with an accuracy of 99.8%. Lastly, I used the unique participants from the face clustering pipeline to design a bias correction procedure that decreased model RMSE by an average of 18% per participant. Technologies used: PyTorch, XGBoost, CuPy, InsightFace, ONNX, NetworkX, OpenCV, Pandas, Numpy, Scipy

Environment Canada and Climate Change (ECCC)

Junior Data Scientist• Sept - Dec 2020

At ECCC, I co-authored a paper detailing the use of machine learning to predict precipitation type, which will be presented at the 2021 General Assembly of the European Geosciences Union (vEGU21). I designed GRU and LSTM models to infer precipitation type from time series atmospheric data, improving classification accuracy from 66% to 91%. I also built a regression model to predict snowfall from time series temperature data, which was applied to 7620 stations across North America. All of this involved writing data cleaning and data visualization scripts for 4 meteorological datasets, each comprising over 1 million data points. Technologies Used: SciKit-Learn, Keras, SciPy, Pandas

PerkinElmer Inc.

Machine Learning Engineer• Jan - April 2020

I worked with a team of 5 to construct a pipeline for identifying the most informative features in high content screening data, with the goal of classifying 4 compounds crucial to drug discovery. I then implemented a deep learning architecture that reduced the number of features from 5000 to 150, while improving model accuracy from 91% to 96%. Lastly, I used model interpretability algorithms to identify biological relationships between features. Technologies Used: SciKit-Learn, Keras, PyTorch, SHAP, Captum

Sunnybrook Research Institute

Medical Biophysics Summer Student• May - August 2019

I co-authored a paper discussing MRI heart models, which was presented to the Society of Cardiac Magnetic Resonance. As part of my project, I developed a GUI that reduced modelling times for finite element heart meshes from 4 hours to 5 minutes. I also used fuzzy C-means clustering to segment scar tissue, with the goal of non-invasively diagnosing heart arrhythmias. Technologies Used: MATLAB, ITK Snap

Skills

I'm experienced in Python programming, particularly with the following frameworks:

  • Python
  • Keras
  • SciKit-Learn
  • Pandas
  • MATLAB
  • Pytorch/FastAI
  • Tensorflow
  • C++
  • C#
  • Git
  • SQL

Client Testimonials

  • John was one of the most dependable coop students we’ve ever had. He consistently delivered excellent quality of work while coming up with creative solutions for feature selection using machine learning. John was extremely driven and self-motivated to research and explore unfamiliar architectures to solve our problem, all with minimal guidance. I was very impressed with his diligence, initiative and his ability to take ownership of his work and I could always count on him to find a solution regardless of the challenges or limitations faced. John was an absolute pleasure to work with and I highly recommend him for any research or engineering role in machine learning or software development.

    Adrian Bar - ML Developer at PerkinElmer Inc

  • John was able to quickly get up to speed and to learn working with data from a different domain knowledge area. He was able independently to find the information he needed to develop and validate deep learning models, and then delivered impressive research results on a tight schedule. I highly recommend him.

    Dominique Brunet - Research Scientist at ECCC

  • John did a fantastic job and exceeded our expectations in almost all aspects. He introduced a new feature selection method that accelerated our model training pipeline significantly, which impressed everyone, he also worked on many other difficult tasks and consistently delivered outstanding work in a timely manner. It's a great pleasure to work with such a smart and diligent young man.

    Paul Zheng - Chief Researcher at Nuralogix

Get In Touch.

Thank you for your time, and I look forward to working with you on future projects.

For business inquiries, contact me at jquinto64@gmail.com or 289-654-4930