Barely-a-CFD Engineer

I had the opportunity to discuss a CFD position at Paanduv Applications with a young applicant who had recently graduated (M.Tech) with a specialization in Computational Fluid Dynamics (CFD). Below are some snippets from that conversation.

Interviewer: Do you have experience in CFD?

Applicant: Yes, I know ANSYS.

Interviewer: Do you have experience with CFD solver of any CFD software?

Applicant: Yes, I can write user-defined functions.

Interviewer: Can you write any simple CFD solver in a programming language?

Applicant: No, but I know Python.

Interviewer: Do you have good fundamentals in numerical methods?

Applicant: Yes, I took a course during my studies, but I do not remember much. I will have to revise.

Interviewer: What other job offers do you currently have?

Applicant: I have a job offer of X lacs from Y company for the position of ML Engineer.

Interviewer: Oh, excellent! So you are an ML Engineer too? What is your motivation to switch to the Data Science field after conducting research in the field of CFD for the last few years?

Applicant: I am getting a good salary in the Data Science position.

Interviewer: Do you have any financial liabilities that you need to address on an urgent basis that you are willing to forgo your hard-earned CFD knowledge?

Applicant: Yes, I have to take care of my family, and then in a couple of years, I have to be married as well. So, a good salary is important to get good proposals for marriage.

What framework do you use to write AI or ML programs?

Applicant: Jupyter notebooks

Interviewer: Can you utilize Tensorflow, Pytorch, or Keras for writing deep-learning code from scratch?

Applicant: No.

Interviewer: Which Linux platform do you prefer for CFD or AI work?

Applicant: Sorry, I do not know Linux, it's too tedious. I do everything in Windows.

Interviewer: What is your expected salary range for this position?

Applicant: Definitely more than X lacs as I am getting that offer at Y company.

Interviewer: Are you willing to relocate?

Applicant: I am getting X lacs already for a remote position. So, I would prefer to work from home.

Interviewer: Well, as the job position states, we are looking for someone with experience in OpenFOAM application and code development. We do have requirements for Deep Learning code developers as well. Do you think your skills or interests match exactly any of these requirements?

Applicant: Not right now, but I can learn if compensated above X lacs.

Interviewer: Thank you for your time. We will discuss this internally and get back to you with whatever decision we make.

This is not just one interview, we have similar conversations with job applicants on a regular basis. CFD graduates are not ready for future CFD applications. They are not ready because they don't see the glamour of sticking to a relatively low-paying CFD job in India and especially when the demand for CFD graduates is of a poor standard. The demand for CFD engineers is basically equivalent to knowing your way around commercial CFD software. Theory, operating system, scientific computing, data analysis, and expertise in CFD software are the five main ingredients that must be looked for in a CFD Engineer. The demand is barely for the last ingredient, the CFD software expertise.

And now on the matter of AI and ML. The CFD resumes are littered with data science jargon such as deep learning, machine learning, Python, Jupyter notebooks, and such, but really no expertise in writing a 200-line Python code utilizing Tensowflow and OOPs concepts. This inability to write code from scratch can be partially attributed to the edTech courses that provide you with pre-written codes and the students have to barely modify (add) a few lines of that code. I understand that some of these courses are designed for short durations, but it's the student's responsibility to learn from there.

Overall. this is a scary situation for a developing economy like India that is preparing a crop of "barely-a-CFD" Engineers. There seems to be huge resistance from top academic institutions and industries to adopting open source CFD. If the concern is accuracy, then companies like us can convince you otherwise. But many times the problem is sheer laziness and the fear of becoming irrelevant in the inevitable mass migration to open source in the coming years.

I understand the concern that CFD jobs are low-paying compared to fancy Data Science jobs, and this article is by no means a motivation to stick to CFD as a career. And I am not trying to lecture about the role that high-paying jobs play in getting "good" life partners in India either. But then, be good at something.

Become a master of at least one trade if not many trades!

This may sound like a marketing article, but I genuinely believe that via our Paanduv Applications Advanced Academic Program (PA3P) we are preparing a good batch of thorough CFD Engineers who are better prepared for the upcoming migration to open-source CFD software

PS: The views here do not endorse any software. And of course, there are a few job applications that are really top-notch and fit well with our requirements.