Licheng Lin
Two years experience delivering academic innovations. Eager to channel my abilities to make a tangible difference in the industry.
Unlimited work authorization.
Education
University of California, Los Angeles (UCLA)
Los Angeles, CA
Master of Engineering, Artificial Intelligence (Computer Science)
Sept. 2024–Expected Dec. 2025
Zhejiang University
Hangzhou, China
Bachlor of Engineering, Information Engineering
Sept. 2022–June 2023
Cumulative GPA: 3.80/4.00, Junior/Senior GPA: 3.88/4.00
Selected Honors: Zhejiang University Scholarship of Academic Excellence, Scholarship of National Talents Training Base.
Experience
Research Intern – University of Miami
Coral Gables, FL
Clustered Federated Learning, Inverse Reinforcement Learning
July 2023–June 2024
  • Advisor: Prof. Mingzhe Chen, Department of Electronic and Computer Engineering
  • Developed an efficient clustered federated learning framework to tackle non-IID challenge of federated learning
Research Intern – Zhejiang University
Hangzhou, China
Secure Federated Learning, Semantic Communications
Sept. 2022–June 2023
  • Advisor: Prof. Zhaohui Yang, College of Information Science & Electronic Engineering
  • Led a team of 3 to design a novel framework for secure federated learning with differential privacy in multi-cell networks
  • Collaborated with a team of 4 to devise a new approach of semantic information extraction for text transmission
Projects and Research
Bookmark ArtChive | Next.js, Tailwind CSS, NestJS, Node.js, NextAuth, Prisma, PostgreSQL | Demo | Github Repo
  • Developed Bookmark ArtChive, a personal art vault software that allows users to save, tag, and rediscover their Pixiv favorites
  • Engineered local download and browsing of Pixiv bookmarks, ensuring uninterrupted access to favorite artworks at anytime
  • Implemented an intuitive organizing and filtering system for users to easily find and enjoy specific pieces in collection
TOEFL SmartBeat | Next.js, Tailwind CSS, Go, Ent, PostgreSQL, AWS S3 | Demo | Github Repo
  • Developed TOEFL SmartBeat, an innovative web platform designed to help users master essential skills for TOEFL preparation
  • Designed a “sentence-by-sentence intensive listening” feature for the listening practice, allowing users to pinpoint challenging sentences and utilize AI-powered repetition to enhance auditory comprehension
  • Leveraged OpenAI's Whisper model to transcribe audio into sentence-level segments, providing accurate practice material
Link Cockpit | Next.js, Tailwind CSS, Flask, Python, MongoEngine, MongoDB | Demo | Github Repo
  • Built Link Cockpit, a personalized command center for swift web navigation, enhancing productivity for frequent users
  • Developed customizable browser start pages for organizing and accessing frequently used links instantly
  • Optimized the application for lightweight and blazing-fast performance, streamlining daily web navigations
Efficient Clustered Federated Learning | IEEE GLOBECOM 2024 | Python, PyTorch | Paper
  • Developed an advanced framework for edge devices with non-IID data to autonomously form clusters and conduct training
  • Enhances clustering efficiency significantly by leveraging local gradient information, optimizing the overall training process
  • Reduces up to 99% clustering iterations, resulting in 14% fewer training iterations without compromising model performance
Federated Learning with Differential Privacy in Multi-cell Networks | IEEE ICC 2023 | Python, CVXPY, CVXOPT | Paper
  • Proposed an secured framework of federated learning with differential privacy in multi-cell networks
  • Optimizes user connections, transmission power and privacy noise jointly by a novel genetic-algorithm-based method
  • Reduces up to 73% total privacy leakage compared to existing algorithms
Semantic Information Extraction for Text Transmission | IEEE/CIC ICCC 2023 | Python, PyTorch, MATLAB | Paper
  • Proposed a new approach for text transmission to semantic information extraction in resource-constrained networks
  • Combines semantic information selection with knowledge graph and probability theory
  • Achieves an additional up to 4x compression efficiency compared to existing methods
Technical Skills
  • Programming Languages – C, Java, Python, HTML, CSS, JavaScript, TypeScript, MATLAB, Verilog HDL
  • Frameworks – PyTorch, Flask, FastAPI, React, Next.js, Tailwind CSS, Node.js, NestJS
  • Tools, Databases, Messaging Frameworks – Git, Docker, Redis, MySQL, MongoDB, PostgreSQL, RabbitMQ, Kafka
  • Familiar with – Spring, Spring Boot, Go, Gin, Gorm, Echo, Rust, Django, Vue 3, Nuxt 3, Google Cloud, OCI, AWS