🧑‍💻Yongzhen Zhao

(+86) 181-6810-0075 | zhaoyzzz@outlook.com

Date of Birth: March 22, 2002 | Place of Origin: Qingyang, Gansu, China

Education

Nanjing University - Master of Engineering - Software Engineering (2024.09-2026.07) (Recommended Admission)

Nanjing University - Bachelor of Engineering - Software Engineering (2020.09-2024.07)

Skills

• Fundamentals: Proficient in foundational knowledge of computer networks, operating systems, and compiler principles.
• Programming Languages: Proficient in Java, Python, and C++.
• Backend Frameworks: Familiar with SpringBoot and capable of quickly learning other Web frameworks.
• Technical Sharing: Shares technical insights on my blog at https://rzyn2020.github.io/.
• Language Proficiency: English (CET-6), capable of proficiently reading English documentation and books, and watching English technical videos.

Internship Experience

Tencent - CSIG - TRTC - Audio/Video SDK (2025.06 - Present)

Tencent Real-Time Communication (TRTC) is a core real-time audio/video communication PaaS product provided by Tencent Cloud. I am primarily responsible for audio/video SDK development.

NetEase - Fuxi - Backend System Development (2023.06 - 2023.09)

AOP (Agent-Oriented Programming) is a new programming paradigm designed by NetEase Fuxi, allowing users to invoke agent services through this framework. I was primarily responsible for developing the DDL module of the project's internal serialization framework. I adopted Test-Driven Development (TDD) and Type-Driven Development to ensure high code quality and maintainability. My main responsibilities included:

• Feature Development: Extended the DDL module's functionality to support various complex data types. Additionally, introduced type hints to the project and integrated mypy for static type checking, effectively improving code readability, maintainability, and type safety, while reducing runtime errors.
• Performance Analysis: Developed performance analysis scripts and utilized flame graphs to pinpoint key performance bottlenecks.
• Test Enhancement: Improved testing for the DDL module based on Property-Based Random Testing principles, increasing test coverage to 90%.
• Performance Optimization: Employed various optimization methods to enhance DDL's performance. Prior to optimization, DDL serialization averaged 50 times slower than protobuf; after optimization, performance was within 2 times of native protobuf in most typical scenarios, and less than one-quarter in some specific scenarios.

Projects

Qiushi Community (2025.04 - 2025.07)

Developed a full-stack, decoupled community platform for philosophy and sociology enthusiasts, integrating AI virtual philosophers for in-depth discussions. I was responsible for technology selection and overall framework design, applying ADD (Architecture Driven Design) and DDD (Domain-Driven Design) methodologies for architectural design and feature implementation.

Tech Stack: Spring Boot, MyBatis-Plus, MySQL, Redis, MongoDB, Elasticsearch, RabbitMQ, Docker

Core Technologies:

• Utilized RabbitMQ to asynchronously decouple user comments, likes, collections, and system messages, significantly enhancing system concurrent processing capability.
• Implemented real-time user activity ranking using Redis ZSET, employing a "write to MySQL first, then delete Redis cache" strategy to effectively ensure cache consistency in high-concurrency scenarios.
• Designed and implemented a distributed ID generation scheme based on the Snowflake algorithm to ensure business ID uniqueness and traceability in high-concurrency scenarios, significantly reducing ID generation latency.
• Integrated RAG and Agent technologies to introduce AI virtual philosophers, enabling them to understand and participate in user philosophical discussions, greatly enriching the community interaction experience.

Transformer-LLM (2025.03 - 2025.04)

• Implemented a Large Language Model based on the Transformer architecture, followed by performance optimization, fine-tuning, and alignment.
• Conducted in-depth research and practiced LLM profiling and performance optimization techniques.
• Through this project, gained a comprehensive understanding of the end-to-end process and key technologies for LLM from model design, training, optimization, to deployment.

SysY-RISCV Compiler (2024.07 - 2024.09)

• Developed a high-performance compiler using C++17, with generated code performance reaching GCC O2 levels.
• Implemented various compiler optimization techniques based on SSA IR, including dead code elimination, constant folding, loop optimization, and register allocation, significantly improving target code execution efficiency.
• Responsible for backend compiler development and served as team leader, enhancing teamwork and leadership skills, and deepening understanding of programming languages.

miniOS Operating System (2022.03 - 2022.06)

• Implemented a multi-processor supported operating system using C language.
• Developed memory allocators based on linked lists, red-black trees, and slab allocation, gaining a profound understanding of the performance trade-offs of different memory management strategies and integrating them into the "Fast Path, Slow Path" system design principle.
• Gained a deep understanding of fundamental concurrency theories and recognized the importance of "defensive programming" in concurrent programming during kernel multithreading implementation.