Jiayin Peng(Eternity)

Jiayin Peng(Eternity)

he/him

ACM Member | Guangdong Key Laboratory for Artificial Intelligence and Multi-Modal Data Processing

Introduction

Jiayin Peng

  • Email:
  • GitHub: EternityPeng
  • Phone: +86 18207572289
  • Language
    • Programming Languages
      • Python,SQL, C++, C, Java, JavaScript, PHP
    • Conversational Languages
      • English: Proficient
      • Mandarin: Native
      • Cantonese: Native
  • Tools & Frameworks
    • Web Development: Flask, Jekyll, Ruby, Node.js
    • Version Control: Git, GitHub, GitLab
    • DevOps: Docker
    • Machine Learning & AI: TensorFlow, PyTorch
    • Database Management: MySQL
    • IDEs & Editors: Visual Studio Code, PyCharm, Clion, IntelliJ IDEA
    • Financial Models & Frameworks: Bayesian Methods, Monte Carlo Simulation
    • Other Tools: LaTeX, Markdown, Bash, Linux Shell Scripting, Mac Shell Scripting, Unity
  • Hobbies & Interests
    • Technology
      • Compiler Construction, Software Engineering, Data Analysis, Internet of Things (IoT), Web Development, App Development, Natural Language Processing (NLP), Computer Vision and Pattern Recognition (CVPR)
    • Sports
      • Table Tennis, Karate, Running, Cycling, Kayaking, Fitness, Archery
    • Other
      • Quantitative Investment, Business & Market Analysis, Chinese Chess, Film, Travel

Summary

  • Completed three years of major courses in two years (transitioned from Film to Computer Science), continuously earning the school’s First-Class Scholarship (top 8% in Computer Science) and Second-Class Scholarship (top 20% in Film).

Education Background

M.Sc., ata-Centric Artificial Intelligence Technology
The Hong Kong University of Science and Technology (Guangzhou)
Graduation: 08.2026 - now

B.Sc., Computer Science and Technology
Beijing Normal-Hong Kong Baptist University (BNBU)
Graduation: 09.2021 - 06.2025

Foshan No. 1 High School Graduation: 09.2018 - 06.2021

Shimen Experimental School Graduation: 09.2015 - 06.2018

Nang Guang International School Graduation: 09.2009 - 06.2015

Honors & Awards

  • National Third Prize in Python Group at the 15th Lan Qiao (Blue Bridge) Cup, 06.2024
  • National Second Prize in the 3rd National University Students Market Research and Business Planning Competition, 06.2024
  • Guangdong Medical Valley Scholarship, 11.2024
  • Third Prize in Interactive Media Design Group at Guangdong University Student Computer Design Competition, 09.2024
  • First-Class Award of BNBU Scholarship, 12.2024
  • First-Class Award of BNBU Scholarship, 11.2023
  • Second-Class Award of BNBU Scholarship, 11.2022
  • Industrial Internet Platform Development Engineer Intermediate Level, 06.2024
  • Second Prize in the 3rd National University Students Data Analysis Science Popularization Competition, 09.2024
  • Market Analysis Researchers LEVEL I, 04.2024
  • LeetCode University Competition Excellent Award (Top 4), 09.2023

Project Experience

The Analyzer of The Nine Chapters on the Mathematical Art (Leader)

02.2024 - 06.2024

  • Reconstructed and optimized the analysis structure of the lexer and parser, utilizing a nested set structure to enhance the semantic analysis phase, particularly the parse tree with type checking, enabling the computation of inverse function equations and solving complex multi-query function problems. This optimization accelerated the dimensionless and dimensional analysis as well as the differentiation and conversion between different basic types, constructed types and other types, resulting in a nearly 70% improvement in compilation speed.
  • Utilized a disjoint set (union-find) structure to enhance the lookup speed of the lexeme table and parse table. Additionally, adjusted the regular expression matching mechanism to accommodate Classical Chinese characteristics, enabling support for converting Classical Chinese numerals, including fractions, to Arabic numerals within a billion range, covering 11 types of units across 97 different measurement units.

A Self-aware Computing Model for Autonomous Driving

03.2024 - 09.2024

  • Multi-dimensional features of driving behavior (such as speed, braking, steering, etc.) are extracted, and the similarity between samples and clusters (such as cosine similarity, Manhattan distance) is calculated. It is used to judge and screen whether there are anomalies in the clusters quantified by vehicles at the same time and route.
  • Collaborated with Assoc. Dean Ricky Hou to clean datasets, subsequently used QGIS to map historical routes and vehicle flow patterns.
  • Agglomerative hierarchical clustering was used to merge the single driving action samples, and gradually formed “single normal action cluster” → “normal driving segment cluster” → “overall normal driving pattern cluster”. In the dendrogram, abnormal driving behavior is defined as those instances that do not merge with any of the normal clusters and form separate branches (e.g., “sudden speed change, sudden braking behavior”), and the hierarchical difference between abnormal and normal behavior is clearly visible. DBSCAN is used to cluster high-frequency driving behaviors such as “smooth driving and normal car-following” into high-density clusters, while low-frequency and rare driving behaviors such as “continuous sudden braking + overspeed + frequent lane change” will be classified as low-density noise points and determined as abnormal driving.

Internship Experience

Canvest SciWin Intelligent Company (HKEX-listed), R\&D Dept.

Data Processing Intern (06.2023 - 09.2023)

  • Performed point cloud localization and annotation using Unity, including classification and statistical analysis of vehicle type, driving status, heading position and other related attributes, and delivered high-quality annotated datasets to support algorithm training.
  • Took charge of the screening, analysis and retrospective review of anomalous data in point cloud annotation (including annotation errors, missed target detection, positioning deviations, etc.). Designed AB Test schemes to compare the accuracy and efficiency of different anomaly detection strategies, compiled full anomaly reports covering data conclusions, root cause analysis of issues and optimization suggestions, and delivered weekly progress reports within the team.
  • Assisted in the installation and deployment of Velodyne VLP-16 LiDAR components, and coordinated with the team to complete equipment commissioning, data acquisition environment setup and parameter calibration.

Activities

  • BNBU ACM Member
  • Guangdong Key Laboratory for Artificial Intelligence and Multi-Modal Data Processing Member
  • BNBU Maker Center “Yanxin” Technical Team Member (First Prize, Innovation and Entrepreneurship Competition; Silver Award, “Jingniuhui” Cup)
  • Technical Advisor and General Head of the BNBU Digital Healthcare Research Group
  • Attend the 8th IEEE International Conference on Energy Internet (ICEI 2024)
  • Behavioral and Decision Research Laboratory Member at United International College
  • Chinese Chess: Third Place (City), Table Tennis: First Place (District), Karate: Second Place (District)
  • Sound Assistant for the No Fear project at the 13th University Student Advertising Art Competition (Second Prize)
  • Presented two research projects at BNBU Poster Exhibition

View more on GitHub