About future

My CVPR submission is fucked. A reviewer straightly said my method was ‘limited both in scientific knowledge advancement and practical deployment’ and gave a 2. But if that’s a case, why not a 1? It feels like the reviewers were neither familiar with the area (i.e. test-time adversarial defense) nor familiar with the paper (since it was probably fed into some next-token predictors anyway).

Anyway, I still think it’s a descent paper, the method was simple yet effective and I will continue submitting it. If it ends up nowhere, then it is what it is.

This experience together with my recent internships at several labs kind of reshape my mind of pursuing a research scientist position in the future. I strongly agree with Weng on the point that it is much easier to teach an engineer to do research than to teach a researcher to do engineering. That said, although I now have certain experience and perhaps some skills of research, there’s still a long way to go.

And as I said before, academia is fucked. This is especially true when you’re actually engaged once in the shitty review system. It either needs a revolution or will crash out first. Another thing is the academia is focusing too much on publishing itself instead of real world impact, at least most institutions do so.

The ideal job for me in my mind now is probably a ‘research engineer’, which means I can apply research mindset to real products and make real impact. Moreover, doing engineering does not mean you cannot make novel ‘academic’ contributions. To achieve this, I’ll need to improve my engineering skills, especially the ability to develop projects at scale. I’ll have to thank RHOS for offering such an opportunity.