Congshu Zou

I am a results-driven computer science professional specializing in machine learning, with hands-on experience in building and fine-tuning models using frameworks like PyTorch, Scikit-learn, and Pandas. I have developed automated pipelines, optimized large language models, and integrated advanced features into graph neural networks. My strong programming skills in Python, C++, and SQL, coupled with leadership, collaborative and analytical abilities, enable me to contribute cross-functional projects effectively. With experience in both academic research and professional environments, I am committed to delivering innovative, data-driven solutions and impactful results.

Education

M.C.S. in Computer Science Specializing in Machine Learning

Concordia University | Mila, Montreal, QC, Canada | Sept 2023 – May 2025

B.S. in Mathematics and Statistics and Computer Science

Concordia University, Montreal, QC, Canada | Sept 2020 - May 2023

B.S. in Chemistry

Fudan University, Shanghai, China | Sept 2008 - June 2012

Projects

Model Merging and Visualization

  • Train models across multiple architectures to evaluate performance and interpretability.
  • Experiment with various model-merging techniques to optimize task-specific outcomes.
  • Visualize intermediate layers to enhance interpretability, contributing to more transparent and explainable AI models.
Accepted Paper: Understanding Permutation-Based Model Merging with Feature Visualizations, UniReps Workshop, NeurIPS 2024

Aligning Language Model with User Search Intent in Query Reformulation

  • Reformulated search queries using large language models (LLMs) to better align with user intent via an AI feedback pipeline.
  • Developed an automated pipeline to rewrite and refine queries with LLMs, enhancing relevance and accuracy.
  • Summarized test results and generated project reports.

Machine Learning for Medical and General Image Classification

  • Developed deep learning models using ResNet-50 architecture to classify medical images as well as animal faces datasets, showcasing the power of transfer learning and pre-trained models.
  • Optimized model performance and interpretability by applying techniques such as transfer learning and t-SNE visualization.
  • Implemented feature extraction with pre-trained CNN models and enhanced classification accuracy using machine learning algorithms such as KNN and SVM.

Enhancing Graph Neural Networks with Chirality-Aware Layers

  • Integrated chirality-aware layers into Graph Neural Network (GNN) models to capture 3D molecular structures.
  • Evaluated model performance on various benchmarks to assess the impact of chirality-aware layers.

Guess a Number Game

  • Developed a browser-based number-guessing game using HTML, CSS, and JavaScript.
  • Implemented responsive and mobile-friendly design with numeric keyboard support and restart functionality.

Play Game

Work Experience

Smardt Chiller Group

Mechanical Consultant Contractor | Jan 2023–June 2024

Mechanical Engineering Team Lead | Jan 2019–Dec 2022

Mechanical Designer | Feb 2014–Dec 2018

  • Led a mechanical engineering team in collaboration with global sales, purchasing, and manufacturing departments to improve internal productivity and deliver high-quality products that met customer requirements.
  • Developed automated Python scripts for generating standardized catalogues, reducing manual work time by 75%.
  • Programmed tools to calculate product weight and dimensions for sales integration, enabling accurate estimation for internal teams and external clients.
  • Analysed historical data and developed a project tracking system using Python and analytics tools, improving team performance monitoring and increasing efficiency by 30%.

FNZ

Junior Software Developer | Nov 2023–Dec 2023

  • Gained hands-on experience with SQL, executing database queries for data extraction and manipulation.
  • Used Jira for task tracking and organization, supporting project management best practices.