Work History
Education
Skills
Honors And Awards
Research And Technical Projects
Timeline
Generic

Junzhe Zong

Work History

Machine Learning Intern

Ligoo Research Institute
Hefei, China
05.2025 - 08.2025
  • Developed a machine learning based early warning system to predict thermal runaway in batteries of electric cars, aiming to prevent hazardous events like explosions.
  • Implemented and evaluated time series analysis models, including a Transformer-based architecture (Sundial), to perform binary classification on battery health status.
  • Engineered predictive and informative features from sensor data to improve model accuracy and robustness.
  • Collaborated with the research and development team to validate model performance against historical data and contributed to the company's ensemble algorithm.

Education

Master of Science - Computer Science

Columbia University
New York, NY
12.2026

Bachelor of Science - Engineering, Data Science

University of Michigan
Ann Arbor, MI
05.2025

Bachelor of Engineering - Electrical and Computer Engineering

Shanghai Jiao Tong University (SJTU)
Shanghai, China
04.2023

Skills

  • Python
  • C
  • Matlab
  • SQL
  • LaTeX
  • Linux
  • Git
  • Bash Scripting
  • PyTorch
  • Pandas
  • Scikit-learn
  • Tensorflow
  • Supervised Learning
  • Data Mining
  • Reinforcement Learning
  • Game Theory

Honors And Awards

  • Summa Cum Laude Graduate, University of Michigan, 05/25
  • ABC scholarship, Shanghai Jiaotong University, 01/23
  • Honorable Mention, Mathematical Contest in Modeling (MCM), 02/22
  • Silver Prize, International Linguistics Olympiad China, 08/20

Research And Technical Projects

  • Multi-Armed Bandit Algorithms with Cost Subsidy, Student Researcher, Prof. Osman Yagan, Carnegie Mellon University, 09/24, 11/24, Pittsburgh, PA, Designed a novel multi-armed bandit algorithm incorporating a cost subsidy parameter to balance exploration-exploitation trade-offs in resource-constrained environments., Leveraged Bayesian Optimization to fine-tune parameters, leading to a significant reduction in cumulative regret compared to classic models., Benchmarked the custom algorithm against established strategies (UCB, Thompson Sampling) on the Multi-Armed Bandit problem, demonstrating superior performance over traditional methods in Reinforcement Learning.
  • Landmark Image Classification Using Deep Learning, Team Leader, Machine Learning Course Project, 12/23, 01/24, Ann Arbor, MI, Led a team to build and train a convolutional neural network (CNN) to classify landmark images with high architectural variance and interclass similarity., Mitigated overfitting by implementing data augmentation techniques (rotation, scaling, flipping) and incorporating dropout layers and regularization., Improved classification accuracy by over 15% by employing transfer learning with a pre-trained model and fine-tuning the final layers.

Timeline

Machine Learning Intern

Ligoo Research Institute
05.2025 - 08.2025

Master of Science - Computer Science

Columbia University

Bachelor of Science - Engineering, Data Science

University of Michigan

Bachelor of Engineering - Electrical and Computer Engineering

Shanghai Jiao Tong University (SJTU)
Junzhe Zong