My Journey: From Information Systems to LLM Engineering
How I transitioned from studying information systems at SHU to working on cutting-edge AI at ByteDance, via data science at USYD.
Shanghai University (2019-2023)
I started my undergraduate journey in Information Systems at Shanghai University. At the time, I didn't know I'd end up working on large language models. Information Systems gave me a broad foundation — databases, software engineering, business analysis, and a systems-level way of thinking about technology.
The turning point was my junior year, when I took a machine learning elective. Building my first neural network felt like discovering a new dimension. I spent the rest of undergrad diving deep into deep learning, reading papers, and building side projects.
University of Sydney (2023-2025)
I chose USYD's Master of Data Science program because of its balance between theory and application. Sydney's program had strong coverage of statistical foundations while also keeping pace with the rapid evolution of AI.
During my time at USYD, the LLM revolution was in full swing. I focused my research on three areas that I found most exciting:
- Large Language Models — Understanding transformer architectures, training dynamics, and emergent capabilities.
- Agent Systems — How to make LLMs that can plan, use tools, and act autonomously.
- Multimodal AI — Bridging vision and language for more complete understanding.
ByteDance (2025-Present)
Joining ByteDance as an LLM Application Algorithm Engineer was the natural next step. The scale of challenges here is unlike anything in academia — millions of users, real-time latency requirements, and the need for systems that work reliably every single time.
My focus on general-purpose Agent systems means I'm constantly navigating the gap between research ideas and production reality. It's challenging, but there's no better place to learn what actually works at scale.
What I've Learned
Looking back, a few things stand out:
- Breadth first, then depth. My IS background gave me systems thinking that pure CS students sometimes lack.
- Follow your curiosity. Every career pivot started with "this is interesting, let me explore it."
- Build things. Papers are great, but shipping code teaches you what papers can't.
The best time to start is always now. The second best time is also now.