English / 简体中文

🤗 About me

Majored in Intelligent Science and Technology at East China University of Science and Technology, be familiar with development and deployment of deep learning applications, keeping up with the evolution of deep learning technology, and trying to develop practical applications in the AI area.

Familiar with Python and its third-party libraries, able to quickly get started with the new framework to achieve the required functions of AI systems.


🔥 News

  • 2024.01: 🎉🎉 Joined TESLA as Service Engineering Intern!
  • 2023.07:  🎉🎉 Won the second prize in 2023 16th Chinese Collegiate Computing Competition!
  • 2023.06:  🎉🎉 Won both first prize and second prize in 25th Chinese Robotic and Artificial Intelligence Competition(CRAIC)
  • 2023.05:  🎉🎉 Achieved an overall score of 7.5 in IELTS!
  • 2023.04:  🎉🎉 The paper was accepted by EI conference CVIDL2023
  • 2023.01:  🎉🎉 Won the second price in 2022 Asia and Pacific Mathematical Contest in Modeling(APMCM)

💻 Projects

Intelligent family tree community

🔥 The possibility of LLM collaboration with humanities and history studies is now being explored! try our demo!

The face recognition technology is used to collect personnel information from group photos, and the tags are combined to associate family social relationships and analyze social networks, so as to realize the visual derivation of community and family tree. Build a end-to-end social/historical humanities research platform.

Pedestrian flow monitoring system based on Deep learning pipeline

widely honored in the national college student competitions(showing below)

Computer vision technologies such as target detection and tracking, image classification, and video classification are used, and intel OpenVINO is combined to realize CPU asynchronous reasoning deployment acceleration, realize good reasoning performance of multi-model pipeline at the edge end, and provide differentiated model deployment solutions between the server and the edge end.

Sports video identification tracking tool set

Cooperate with the national climbing team for the live broadcast of the event

Real-time sports event recognition and tracking toolset based on PP-Human, a real-time pedestrian analysis tool using the paddlepaddle deep learning framework, integrating various deep learning models to achieve customized functions (player tracking, football possession detection, ski posture, player highlighting, ball flight trajectory fitting, etc.)

  • Technical route: PaddlePaddle+OpenCV+TensorRT
  • Code: Github

📝 Publications

CVIDL2023
sym

Pedestrian flow monitoring system based on Deep learning pipeline

Yifan Bu

  • Designed a Pedestrian flow monitoring system, which uses a variety of deep learning technologies (including target detection, target tracking, video classification, etc.)
  • Three different solutions for different traffic scenarios, covering heat maps and target tracking
  • Built a system composed of deep learning pipeline, data visualization and database to achieve the purpose of monitoring the flow of people in crowded public places
CIPA2023
sym

RESEARCH ON THE TRANSFORMATION OF HISTORIC PATTERNS OF OLD SUMMER RESORTS USING SOCIAL NETWORK ANALYSIS: A CASE STUDY OF KULIANG IN FUZHOU, CHINA

Y. N. Lin, C. Yang, Y. X. Chen, Y. F. Bu, C. L. Ping, and B. Y. Cheng

  • The seemingly randomized distribution of buildings and historical landscapes in Kuliang is a result of social relationship development.
  • The formation of Kuliang’s summer resort is correlated with the selection of “central members” in the community and has distinct temporal characteristics.
  • Information technology has great potential for the analysis of the formation process of cultural landscapes. These research findings provide important reference for the protection of cultural heritage and understanding the role of social relationships in shaping cultural landscapes.

🏢 Professional Experience

SangFor Technologies Inc - embodied ai devices

Backend development Engineer

2024/07/08-now

  • Development of the Data Management Platform using Golang and Python, integrating the entire workflow from data collection, annotation, visualization, to export processes. Addressing data management shortcomings in open-source solutions.

  • Development of a Cluster Visualization Monitoring Platform, providing visual representation of container-level GPU utilization. Utilizing Grafana and OpenTelemetry for the collection of microservice traces to diagnose and troubleshoot live service issues.

Tesla(Shanghai)Ltd

Service Engineering Intern

2024/01/24-2024/07/06

Certification

  • Responsible for the development and maintenance of the internal K8S data platform, using machine learning algorithms to efficiently diagnose potential dangers from massive IoT data and deploy it to production cluster.

  • Optimize performance in a variety of ways to reduce task time by more than 30 times while maintaining the original functionality.

  • Develop a data storage and encryption platform for accident proof(To G)

China Innovation Center of Roche

AI Development Intern of the AIDD (Artificial Intelligence, Data and Digital) Dep.

2023/07/03-2024/01/19

Certification

  • Assisted in developing a system that utilises computer vision technologies to extract and identify chemical formulas from patent data and realize the digitization and automation of preliminary research
  • Fully engaged in the back-end development for the mentioned system’s implementation workflow which involved the segmentation/recognition of chemical structure molecular formulas in PDF images, the extraction of fingerprints for nearly 2000 molecular formulas, and the dimension reduction/clustering/visualization of data.
  • Built (with Redis and Celery) a concurrent system containing 7 workers and 5 queues/priority queues on the Linux server to achieve fair allocation and scheduling of task resources under high concurrency conditions
  • Encapsulated PyTorch and its essential CUDA environment with docker containers to simplify the migration between platforms
  • Used the PostgreSQL database to manage the massive structural molecular data extracted from patents, providing data support for subsequent machine learning statistical analysis.

Continental Automotive Systems (Shanghai) Co., Ltd

Software Intern

2023/01/09-2023/05/08

Certification

  • Researched speech recognition datasets for academic use, tested them on the platforms of major voice service providers (such as Google, Azure and ByteDance) as well as some open source speech recognition projects (such as Whisper and PaddleSpeech), and produced reports on their word error rate (WER) and char error rate (CER)
  • Screened out suitable speech recognition/speech synthesis service providers (online and offline) and developed the demo of a voice-based intelligent assistant by integrating large language models (LLM) like ChatGPT, ChatGLM and Alpaca
  • Investigated the practical applications of the LLM models ChatGPT/ChatGLM/Alpaca and built up a local knowledge-based natural language document Q&A tool through the Prompt Engineering library lang-chain

🏆 Honors and Awards

I’m so sorry that I may can’t translate all these awards into English correctly. However, all awards mentioned below are commonly honored among Chinese Colleges.


📖 Educations

  • 2020.09 - 2024.06, East China University of Science and Technology,Intelligent Science and Technology (major degree)
  • 2020.09 - 2024.06, East China University of Science and Technology,Information and Computing Science (minor degree)

💬 Invited Talks

  • 2023.02 Pedestrian flow monitoring system - Visualization based on AI makes the city safer - PaddlePaddle Development Expert “Yifan Bu” and Intel AI software engineer “Yicheng Yang”
  • 2022.09-2023.09, PaddlePaddle Development Expert——Yifan Bu