Ann Arbor Machine Learning

PmvoUTRn************

Education

College of Engineering, University of Michigan

M.S.E in Electrical and Computer Engineering

2024.09 - 2025.08

Ann Arbor, MI

College of LSA, University of Michigan

B.S. in Computer Science & Psychology - GPA 3.75/4.0 2020.08 - 2024.05

Ann Arbor, MI

2020 Dean's List, 2021 Dean's List, 2022 Hub Contribution Awards, 2023 University Honors Skills

Full Stack Dev: React, MySQL, AWS, Swift, TypeScript, Linux, ROS, Docker, Flask, Swift, Python, C++, Rust, JavaScript ML & AI: NumPy, PyTorch, TensorFlow, MATLAB, HuggingFace, MMengine, ONNX, OpenCV Work Experience

Vision-Language Model Development Intern

Machine Learning Engineer Confidential Company (due to NDA) 2024.05 - Present

Vision-Language model development: Developed a fine-grained Vision-Language model for food inspection, enhancing CLIP modality alignment in embedding space and increasing accuracy by 10%. Data collaboration and augmentation: Collaborated with data and full-stack teams to fetch high-quality datasets and efficiently generated additional datasets, boosting model performance by 15%. Video Representation Learning Research Intern Paper submitted to NIPS Research Assistant College of Engineering

2024.01 - Present

Ann Arbor, MI

Action recognition model: Assisted in developing an action recognition model, improving downstream tasks by 20%, and conducted an ablation study on action recognition and localization with Faster R-CNN. Algorithm design and hyperparameter tuning: Designed an innovative random selection algorithm for a dual-way CNN, reducing overfitting by 25%, and fine-tuned model hyperparameters to increase mAP by 20%. Experiments and contributions: Replicated contrastive learning experiments on various datasets, and submitted a PR to MMAction2 that accelerated AVA dataset preparation, reducing time by 30%. E-Bike ADAS System Design Co-op

Team Lead Drover.AI

2024.01 - 2024.05

Ann Arbor, MI

ADAS development: Led a team in creating a $100 ADAS for e-bikes, achieving a $15 cost reduction and 30% performance improvement by researching SBC solutions. Model evaluation and deployment: Wrote a custom dataloader for evaluating models on the nuImages dataset and deployed the model in ONNX format for seamless SBC integration. Algorithm and library enhancements: Designed and implemented collision detection algorithms with OpenCV and improved the torchvision video clips library, reducing load time by 10%. Multi-Modality AI Research

Research Assistant College of Engineering

2023.08 - 2024.04

Ann Arbor, MI

Docker and API development: Constructed a Docker container to enhance multi-modality understanding and image generation, and utilized Flask to build an API for generating cross-modality captions. Algorithm improvements: Improved algorithms for computing similarities between modalities, increasing precision by 15%, and implemented a new algorithm from the literature, reducing computation time by 25%. Interface design: Designed an easy-to-use interface with React and TypeScript to streamline the image generation process.

AWS Platform Development Intern

Technical Team Lead Blue Talks

2023.03 - 2023.07

Ann Arbor, MI

Platform development: Engineered a chat platform using AWS services, incorporating a SAML authentication system with the University of Michigan as IDP for 1200 requests. Chat support: Supported call-in and real-time chat bots with AWS Connect, managing 70+ calls and 500+ chats. Front-end development: Developed a dynamic front-end interface using React and TypeScript. Project Experience

Mobile App Dev Projects

Course Developer College of Engineering

2023.09 - 2023.12

Ann Arbor, MI

Course development: Helped a professor R&D a new mobile development course, leading a team of 4 and utilizing AWS EC2, Nginx, and Gunicorn to architect an efficient backend capable of handling 2000 mixed type queries. Video streaming and interface development: Leveraged HLS to implement video streaming capabilities, providing a dynamic user experience, and developed a user-friendly interface using Swift to significantly improve the app's usability.

WebRTC implementation: Constructed a complete suite of WebRTC in iOS, enabling P2P video chats across platforms. Zesen Zhao(Hyman)

+1-814-***-**** ********@*****.***

Ann Arbor, MI

Linkedin: www.linkedin.com/in/zesen-zhao-b1b859244

Contact this candidate