Yongtai Chen

Master of Science in Computer Science

Bachelor of Engineering in Computer Science

• Smart City Platform Design and Implement, Software Engineer Intern
• Created an automated building management platform utilizing Python and Java, and implemented standardized access for multiple manufacturer’s home appliances via REST API
• Implemented high-performance asynchronous event handler using Python’s asyncio module
• Managed and stored diverse data utilizing PostgreSQL and MongoDB databases
• Decreased energy consumption by 30% through conducting user behavior analysis and optimizing lighting and acoustic settings.

• Store Tag Detection and Recognition, Software Engineer Intern
• Link github.com/Acytoo/baidu_ai_competition
• Developed a neural network for detecting storefront billboards using Python and Baidu PaddlePaddle framework, resulting in 2nd highest precision among 21 contestants
• Implemented the Single Shot MultiBox Detector (SSD) neural network, improving upon traditional Convolutional Neural Networks (CNN) with reduced computational requirements
• Deployed the model on Baidu AI Studio, tuning and testing the SSD to optimize performance
• Enhanced speed and accuracy of information storage for Baidu Map, serving 20 million+ daily active users in China

• Deployed and maintained websites, including acytoo.com, illusion.acytoo.com, zhaoge.org, asknyu.com, and yongtaichen.com on Kubernetes, ensuring seamless and efficient operation
• Integrated ChatGPT into the websites, enhancing user experiences and providing personalized assistance
• Leveraged various technologies such as Docker, MariaDB, PostgreSQL, and MongoDB to support website deployment and management

• Conducted research on identifying misinformation about Covid-19 using Python and Spark
• Utilized PySpark.ml to pre-process tweet data and employed GloVe 100D and Universal Sentence Encoder to map cleaned tweets to vector space
• Trained the model using DNN from Spark NLP, resulting in a 91.89% F1-score
• Implemented Spark-Streaming to detect fake news in real-time

• Link github.com/Acytoo/Reverse-Visual-Search
• Designed an image search engine, implemented with Python and TensorFlow
• Detected and cropped the facial part of images using multiple pre-trained neural networks
• Constructed K-nearest neighbors indexes for images in a target dataset based on the extracted features
• Searched the top similar images for the target image using KNN, and achieved 85% accuracy on the LFW dataset

• Link github.com/Acytoo/Kalman-Filter-Multiple-Object-Tracking
• Conducted research on object tracking in a two-dimensional world with multiple objects and obstacles
• Developed a Python-based tracker using the Kalman filter algorithm
• Automated the assignment of multiple trackers to each object
• Achieved accurate tracking even when similar objects crossed over each other

• Link github.com/Acytoo/salient-object-detection
• Detected salient objects without prior knowledge by calculating regional contrast values, implemented the algorithm using C++ and OpenCV
• Improved calculation speed by quantizing into 12 colors and computing pixel saliency in the CIELAB color space
• Optimized pre-processing tasks for various computer vision applications to enhance efficiency on large industrial datasets
• Benchmarked and increased speed by 100% compared to the previous study without losing accuracy on the MSRA10K dataset, processing over 80 images per second on a quad-core 2.7 GHz machine with 8 GB RAM
• Awarded the Excellent Graduation Project by Northeastern University

• Link github.com/Acytoo/File_system_simulation
• Developed a file system simulator with inode-block structure using C++
• Implemented Linux file system calls including file copy (cp), move (mv), and access control (chmod)
• Created a custom data-serialization format to enable file encryption, with support for zip and unzip operations
• Utilized the Qt framework to build a comprehensive UI/UX for end-users, supporting both Linux and Windows platforms

• Link github.com/Acytoo/NewHpClient
• Led a team of four to design and develope a Smart Calendar system using Java, facilitating efficient communication between faculty members and students by real-time notifications
• Applied Agile development methodologies to ensure efficient and top-notch project delivery
• Implemented cross-domain cookie persistence for seamless user synchronization with the school’s message board
• Reduced bandwidth usage and enabled server-initiated data transmission by WebSocket connections built with OkHttp