Determining My Early Career Direction

Mar 12 2026

What AI Won’t Replace

When I started using vibe coding, I realized something profound was happening. No wonder so many PMs keep saying they’ll replace programmers—I’ve actually seen colleagues with zero experience ship web apps.

So where does that leave us CS graduates?

AI suggests focusing on complex tasks or domain-specific work. I’m not particularly interested in business logic, so let’s talk about tackling complexity instead. Think distributed systems, high concurrency, load balancing. I’m not entirely sure yet, but heading toward systems that more people use and that sit at the core of infrastructure seems like the right call. I need to consciously steer my work in this direction.

Small-scale software development and frontend work feel increasingly pointless.

Infrastructure

I’ve suddenly become obsessed with infrastructure. I like applying the barbell strategy to engineering work with high certainty.

It’s a massive field, so I had AI help me categorize it (see the table at the end).

Personally, I feel infrastructure engineering is deeply technical, and the systems you build need to be genuinely complex. However, I don’t want to go deep into hardware—I don’t have much foundation there. AI recommends roles like AI Infrastructure Engineer, ML Platform Engineer, DevOps Engineer, Cloud Architect, or AI Operations Engineer.

Plus, infrastructure skills transfer well from tech companies to quant dev roles. I just need solid financial markets knowledge, which is exactly what I’m planning to spend Saturdays learning.

Current Focus

Learning complex applications while looking for opportunities to work on complex system architecture.

Dedicating an hour daily to personal project preparation.

Stay Open

Keep an open mind.

Sometimes interesting new products—like browsers or VS Code—are happy accidents. You can’t plan everything rationally; you need intuition too. Everything is change.

Apr 12 2026

Understanding about AI Infra

I’m currently most interested in AI infrastructure. In my view, AI infrastructure is about maximizing the capabilities of underlying hardware and systems, while building a solid foundation that enables higher-level software to operate efficiently. As AI continues to develop, more industries will rely on AI-driven software, which makes infrastructure—the backbone of AI—both critical and promising.

In addition, I’m more inclined toward engineering than research. Ideally, I would like to spend around 80% of my time on engineering work and 20% on exploring and learning new technologies.

Of course, there may be more cutting-edge and exciting areas than AI infrastructure. However, as a junior, I believe it’s reasonable to stay focused on AI infrastructure while keeping an open mind toward new and evolving engineering fields.

I really hope we can use local llm, cause I am too worry about openai and claude code have too many power. With the development of opensource llm and hardware, I hope we can do that.

Jul 5 2026

HPC, Distributed programming, Parallel programming.