Salary.sg Forums - View Single Post - Salaries of Professors in SG
View Single Post
  #85 (permalink)  
Old 21-01-2014, 03:13 PM
lordlad lordlad is offline
Senior Member
 
Join Date: Oct 2013
Posts: 50
lordlad is on a distinguished road
Default

Quote:
Originally Posted by Donny View Post
All right, fair enough. Somewhat do agree with what you said.

I guess it goes down to the level of understanding each candidate can bring to the job. I don't deny that those Vietnamese programmers you hire can get some jobs yet.

Yet, I feel that a Stanford PhD (since that is what the OP is aiming at) can bring in new knowledge that produces a about a better solution viz-a-viz those done by the programmers you mentioned.

Right of the top of my head and I'm not trying to show off or be verbose here:

1. Parallel programming on a GPU (undergrad would at best take a single course on GPU programming. PhDs develop and understand better the theory)

2. Machine Learning for signal detection (a PhD would throughout the course of his thesis applied machine learning to numerous problems. An undergrad probably just learnt PCM* and did a two week project on it)

3. Use of appropriate algorithms (perhaps here the PhD and undergrad can be on par. Nonetheless, doing and using algorithms for two years compared to two semesters inevitably suggest more familiarity with it for the PhD than the undergrad)

My point is that a bachelors degree equips one with fundamental knowledge, a PhD with specific knowledge in a specific area. Depending on the job, sometimes both the bachelors and PhD survive, sometimes only the PhD survives. I can trust a Vietnamese programmer to write a multithreaded console program to read and display text from a server to a screen. I won't trust a Vietnamese programmer with at most one year of experience, to tear up a Linux kernel and write a customer memory manager taken from some IBM publication written in the 1980s. (I've seen some of those, they scare the freak out of me.)

*Principle Component Analysis
GPGPU (or GPU Compute) is still a relative 'young' industry (at least in its application).....Companies like Nvidia are currently doing it and pushing it though with their Nvidia GPU based server racks...its application probably more for data crunching, catastrophic financial modeling.

The need for that is still young in the industry at the moment but as it's still emerging, i do feel it's good to start now.
Reply With Quote