It’s been a while since I last updated this blog. Things have been pretty hectic lately, so I figured I’d take advantage of some free time today to jot down what’s been going on — a checkpoint of sorts.

Paper Accepted

The paper I submitted to ICIC got accepted.

From the initial experiment design and data processing, through several methodology pivots along the way, to post-review revisions — it was quite a journey. The biggest takeaway isn’t the acceptance itself, but the methodological instincts I built up through the process: what makes an experiment design solid, which metrics are actually convincing, and how to diagnose and redesign experiments when results don’t match expectations. That kind of know-how is worth far more than any single paper.

---

Work: An AI Project Going Into Production

My company is currently pushing forward an AI project — building an AI Agent solution for an automotive OEM. The scope covers everything from developer productivity, test automation, to overall engineering workflow optimization.

Projects like this are fascinating because, unlike pure research, you can’t stop at the demo stage — it has to actually work in production, deliver real value, and integrate seamlessly with existing toolchains, processes, and people. Along the way, you get a front-row seat to the boundaries and limitations of LLMs in real enterprise settings. These are things you simply can’t learn from papers — you have to hit the walls yourself a few times.

---

Self-Studying Quantitative Trading

I’ve recently started learning the fundamentals of quantitative trading. The plan is to spend a few months going through it properly — from basic financial market concepts, to common strategy types, to backtesting frameworks and risk management tools.

The motivation is twofold: partly personal curiosity, partly wanting to see if my background in ML and data can transfer to a different domain. I’m not expecting to make money in the short term — it’s more about building a long-term knowledge framework and getting some hands-on practice along the way.

---

Post-Quantum Cryptography × Automotive V2X

Another direction I’m working on is the application of Post-Quantum Cryptography (PQC) in automotive V2X communication.

This area is both cutting-edge and highly practical: as quantum computing capabilities advance, the signature algorithms currently used in V2X face a real risk of being broken in the medium to long term. Meanwhile, automotive products have exceptionally long lifecycles — a car stays on the road for well over a decade — which makes PQC migration an unavoidable engineering challenge. I’m currently surveying the performance, bandwidth overhead, and integration complexity of several mainstream PQC signature schemes in vehicular network scenarios. I might put together a dedicated write-up later.

---

Final Thoughts

That’s about it. There’s a lot going on and the pace is pretty intense, but thankfully it’s all in areas I’m genuinely interested in, which keeps the energy up.

I’m hoping to get the blog cadence back on track — too many ideas fade into vague impressions if you don’t write them down in time. That’s a shame.

See you next time.