Fortran is easy and quick to learn, but very hard to master. This may seem contradictory, but think of it as learning to speak and writing, and being able to write like Shakespeare.
The thing is that Fortran's strength is its number crunching capability and efficiency in that respect, and the influence the program's structure has on the final speed. For instance, with large multi-dimensional matrices, accessing elements in the wrong order may reduce the speed by order of magnitude by causing frequent load from the main computer memory to the cache, while accessing elements in the exact way they are stored will restore performance.
Python is an interpreted language. MATLAB is a commercial tool. They may get you there, but while you may save on development time, your final program will be slow to execute. What is needed here? A quick one-off answer, or massive amount of runs for lots of data processing? If he former, got with Python or Matlab. If the later, Fortran is still the king of the hill.
As for computer science guys thinking Fortran users are dinosaurs, well, their perception is flawed and colored by what they teach these days; I presume those IT folks are in their 20's, right? Then they've never really been exposed to Fortran. And if you think about it, 15 years ago, people were nuts over C. Then it was C++. Then .Net. Or Java. See a trend here? The languages come and go.
But Fortran has such a strong library of time proven codes -- some are legacy, perhaps even archaic, but they WORK -- that it has to remain.
I still can write Fortran code (for number crunching--Fortran is NOT for system programming and interface and windows, the same way a Formula 1 is not meant to deliver piano--that will out perform anything anyone can write in any other language. That is why I still code in Fortran.