The pain of an inexperienced "Citizen" MLOps

I had an interview and when I was asked what tools I used. I responded by Using unified platforms like Python, Amazon sage, vertex AI, data bricks, for ML.

I was also asked how do I bring in data from other sources. I responded with Data fusion, Trifecta.

And I was told that I won’t be considered because I do not use apache beam and Kubeflow.

I am still a bit confused.
I thought the end of the goal was to solve a problem irrespective of the tool or software.

Is there a reason why I have to go through a tedious and time-consuming process when I can solve the problem with a managed service?

I thought having theoretical knowledge of solution workflow and executing it in your preferred platform or tool is supposed to be how it works.

Well, what do I know as a 1 year experienced AI personnel?

3 Likes

That sounds awful. Im sorry. If the interviewers dont understand how your experience is sufficient to easily adapt to alternative software products if required, maybe this business isnt a good place to work anyway - (or perhaps they just need someone who can ‘hit the ground running’ ). Sometimes it takes a few applications to succeed so keep trying !

2 Likes

Hi, @Ayoola, and welcome to the community! :wave:

In general, I agree with you. What matters first is your ability to solve the problem.

But, if the team you are applying to is dedicated to using a particular toolchain and interviewed a candidate, who is similar to yourself and familiar with the toolchain, they may pick them simply because they are likely to spend less time learning the toolchain.

Don’t get discouraged. There are companies out there that are a better fit for you :slight_smile:

This is the question I would’ve asked the interviewer. I can imagine workflows where it can be tricky or impossible to use a managed service. But, it depends a lot on how the team uses them, and I don’t want to speculate.

1 Like

From my point of view, this is not your need @Ayoola , because there are a lot of cloud services and ML pipelines. If you not applying as a devOps engineer, you don’t need to know all these services and pipelines. I think they are expecting 1-year experienced AI personnel to do choose service and design pipeline. So may be this is not a good place to work for 1-year experienced AI personnel.