Curriculum Vitae

Education

Technical University of Berlin

M.Sc. in Computer Engineering (3.0/4.0)

2025/04 – 2027/03

Berlin, Germany

University of Erlangen-Nuremberg

M.Sc. in Medical Engineering, Grade: 1.8 (3.2/4.0)

Teaching Assistant of Biomedical Signal Analysis

2023/10 - 2025/03

Erlangen, Germany (Transferred from)

Shanghai Jiao Tong University

University of Michigan – Shanghai Jiao Tong University Joint Institute

B.Eng. in Electrical and Computer Engineering, GPA: 2.7/4.0

Minor in Data Science. Degree taught entirely in English.

2017/09 - 2023/06

Shanghai, China

Experience

ITER Organization

AI Engineering Intern

2025/09 – 2026/02

Saint-Paul-lez-Durance, France

  • Real-time document type suggestion: Engineered a high-throughput inference service for document classification (50+ types) using FastAPI, ModernBERT on vLLM and vector search. Achieved >95% Top-3 Micro AP while maintaining <3s latency under heavy load (30 concurrent requests, ~700 tokens/req).
  • Searching large document bases: Prototyped and delivered agilely multiple searching services using Streamlit and vector database on heterogeneous data sources, implementing semantic reranking and LLM scoring.
  • Event-Driven AI Agent for IT Support: Architecting a scalable Chain-of-Agents system using LangGraph and RabbitMQ to orchestrate Jira ticket resolution. Integrated SAP, Jira, and Confluence MCPs for context-aware reasoning, automating resolution paths, and suggestions for 200+ tickets daily.
  • Introduced LangGraph, Taskfile, uv & ruff, and pytest to the team's workflow.
  • Internship evaluation: 59/60

University of Erlangen-Nuremberg

Teaching Assistant, Biomedical Signal Processing

2024/10 – 2025/12

Erlangen, Germany

  • Maintained automated homework judging system with Github Classroom
  • Crafted exams for 3 semesters

ANKI Lab, University of Erlangen-Nuremberg

Research Assistant

2024/04 – 2024/07

Germany

  • Designed and implemented a biomedical video segmentation workflow with PyTorch

TuringQ Co. Ltd.

System Administration Intern

2021/12 – 2022/02

Shanghai, China

  • Managed user accounts and permissions with role-based access control

Shanghai SmartState Co. Ltd.

Product Management Intern

2021/05 – 2021/08

Shanghai, China

  • Conducted market analysis and procurement of 3D CAD software for a novel GenAI product (generating CNC file)

Wireless Networking & AI Lab, Shanghai Jiao Tong University

Research Assistant

2017/10 – 2019/05

Shanghai, China

  • Designed data cleansing and feature engineering workflows for smartphone power consumption data
  • Created data analysis pipelines based on unsupervised-learning algorithms with Sklearn

Personal Projects

Individual Developer

2025/12 – Present

Open-Source Contributions:

  • Actively contributing to high-profile Rust projects including nushell and uutils/coreutils, focusing on bug fix

System Programming with Rust:

  • Implementing a SQLite clone from scratch to master low-level storage engines, B-tree indexing, and memory management in Rust.

TaggerNews

Personal Project

2025

  • HackerNews digest website with dynamic content tagging using LLM-based multi-agent chains
  • Migrating to a from-scratch C++ web server with custom EventLoop and thread pool, with CI/CD

Fordex Code / GResearch

Personal Project

2025

  • Code: Agent orchestration & communication protocol to enable multi-developer-agent cooperation with native Claude Code Agent Team support
  • GResearch: A DAG-based MCP server that serves as a stateful shared brain across Claude Desktop and Claude Code sessions

LoRA merging for Continual Learning of Pretrained Vision Transformer

Research Project, RSiM Lab, BIFOLD Berlin

2025/04 – 2025/08

  • Integrated LoRA into SpectralGPT (ViT) to enable multi-modal (Spectral/Visual) processing for remote sensing tasks.
  • Conducted performance benchmarking on high-performance computing (HPC) clusters to evaluate model scalability.
  • Implemented LoRA merging strategies to mitigate catastrophic forgetting in Continual Learning, achieving 50% macroAP retention.

Feed-Forward Network (FFN) with CUDA for Data and Model Parallelism

Course Project, Multi-core Architecture and Programming, University of Erlangen-Nuremberg

2024/04 - 2024/07

  • Developed a FFN from scratch using C++/CUDA, implementing custom kernels without external libraries
  • Architected and compared Data Parallelism (DP) and Model Parallelism (MP) schemas with CUDA Stream to emulate parallel computing on single GPU, optimizing training throughput
  • Benchmarked their performance on MNIST, achieving 30% training speedup for DP with no accuracy degradation and 60% for MP with minimal accuracy degradation

BERT on ARM: Training on Server and Real-time Inference on Edge

Bachelor Thesis, Efficient and High Performance Computing Lab, Shanghai Jiao Tong University

2021/09 - 2022/01

  • Orchestrated a computational offloading schema for Huggingface transformers to balance compute loads between cloud servers and NVIDIA Jetson edge devices
  • Optimized transformer inference on ARM architectures, achieving real-time latency for Question Answering tasks

Skills

Programming Languages: C++, Python, Rust (learning)
Web & Backend: FastAPI, Streamlit, PostgreSQL
ML/AI: PyTorch, vLLM, LangGraph, HuggingFace
Infrastructure: Docker, Git, Azure, Google Cloud, Kubespray, Slurm, CUDA
Languages: English (TOEFL 110, C1), German (C1), Mandarin (Native), French (A2)