Zhewen Shen

đź“§[email protected]
GitHub: @zhewenshen LinkedIn: @zhewenshen

Education

UNSW Sydney
Bachelor, Advanced Computer Science (Honours)
  • Honours Class 1 and The University Medal

  • Thesis: Evaluating Programming Languages for Verified LionsOS Components

    • Supervised by Scientia Professor Gernot Heiser

Work

Arista Networks
Software Engineer
  • Implemented anti-rollback protection for Arista EOS

  • Brought up BIOS and bootloader support for products based on Intel Amston Lake platforms

  • Engineered CPU–NIC protection by leveraging hardware-based packet filtering and extending kernel drivers to sustain line-rate 64-byte Ethernet traffic—14.88 Mpps at 10 GbE and 37.20 Mpps at 25 GbE

  • Triaged software defects and customer escalations, diagnosing issues across EOS, the Linux kernel, device drivers, and platform firmware

School of Computer Science and Engineering, UNSW Sydney
Casual Academic
  • Teach classes for COMP6771 Advanced C++ Programming and COMP9242 Advanced Operating Systems

  • Guide students through systems programming concepts, modern C++ features, and OS-level design principles

  • Provide assignment feedback and support students during consultation hours

School of Computer Science and Engineering, UNSW Sydney
Research Intern (Trustworthy Systems)
  • Contributed to LionsOS, a component-based operating system built on the seL4 verified microkernel, designed for building custom, task-specific systems from composable components

  • Implemented core migration and offlining mechanisms for multicore management in LionsOS

  • Built a userspace webserver as a LionsOS component using its lock-free queue-based communication model

  • Created a proof-of-concept transpiler to convert C code into Pancake, a research language with verified compilation for writing formally verified OS components

  • Co-authored a paper at the Workshop on Programming Languages and Operating Systems (PLOS) 2026

School of Computer Science and Engineering, UNSW Sydney
Research Intern (NLP)
  • Developed BAMBINO-LM, a continual pre-training approach for small-scale language models inspired by bilingual language acquisition in children

  • Combined language alternation strategies with PPO-based perplexity rewards to enable cross-lingual transfer between English and Italian

  • Published at the Workshop on Cognitive Modeling and Computational Linguistics (CMCL) 2024

Skills

Languages: Advanced
Keywords: C, C++, Python, Bash
Tools & Infrastructure: Intermediate
Keywords: Git, Perforce, Docker, GDB, Make, CMake, Linux, MongoDB

Awards

The Jump Trading Hero of Operating Systems Prize
UNSW Sydney
The Optiver COMP6771 Advanced C++ Programming Prize
UNSW Sydney
Taste of Research Scholarship (2024 & 2025)
UNSW Sydney Faculty of Engineering

Publications

BAMBINO-LM: (Bilingual-)Human-Inspired Continual Pre-training of BabyLM
Workshop on Cognitive Modeling and Computational Linguistics (CMCL)