This time I decided to write something from my experiences. More precisely, I will give you reasons to refuse that data science offer you got.

I graduated in 2019, with a degree in Chemical Engineering. I had been doing data science for a while. I think I started taking it seriously in 2018 summers during my internship.

Before my internship, I had an immense interest in management consulting. The company where I interned put me on an NLP project. I hardly knew python before that internship. But I had amazing peers who helped me.

After my internship, I was offered a full-time role at the company. I had learned from random places that ‘You can take risks early on in your career’ Join startups. Startups have a great learning curve.’

So I started looking for jobs at startups. Interviewed with a few. In the rest of the post, I will share how to learn to evaluate startups and not be naive. These are based on my experiences.

Easy interviews - sweet poison

If you feel that you nailed your data science interview, ask yourselves - Is it because the interview was about something you were really good at, or is it because the interview was objectively easy?

If the first interview was easy, wait for the second round. Ideally, the second round should be more difficult. If the subsequent rounds are too high-level and easy and you end up with an offer - RUN. As quickly as you can.

It only implies that the team you’d be working with are not experienced and knowledgable. Or, they don’t have good Data Science projects. Or, they have no set data science team.

Who is the senior most Data Scientist?

Make sure you stalk the hell out of everyone in the company on LinkedIn. If you can’t find the team, ASK the interviewer for their LinkedIn profiles. I actually did that. A good company will be impressed by the initiative you took. Any decent startup understands that a candidate needs to do extra research on a relatively unknown company.

Get an understanding of the kind of experiences the team has had in their previous roles. How experienced is the senior-most Data Science/ML Engineer? Ideally, 5+ years is preferable. However, previous experiences should be fairly relevant to their current job.

Beware of the visionaries

If you are interviewing with an early-stage startup, chances are that you will speak to the CEO. If the CEO mentions any of these phrases

We have no competitors. We are unique and no one else is doing what we are.
Every other company doing stuff similar to what we do is shit. They don’t know what they are doing.
You are given a chance to be a millionaire.
With this equity percentage, you will be $$ millionaires $$ in 2 years. We can’t pay you more salary though.
Our idea is life-changing

BEWARE!!

A good CEO is practical and acknowledges the competition. In one of the interviews, I asked the CEO - "AWS has the same product as you. What if they surpass your Machine Learning’s accuracy? “ He calmly said -

Sure. Please go ahead and use their services.

Not a single attempt to bullshit. It is okay for a founder to not know the answer to every what if question. But one’s who lure you into believing random stories, beware of those.

How good is the software team?

Remember that data scientists job exists because there are engineers. They are the ones who integrate your model into the product. If there are no good engineers in the team, RUN.

Most probably, the startup is fooling you by offering you the role of Data Scientist but would actually make you do software development all the time.

Remember that most startups need an excellent software team way before data scientists. Any technically sound founder knows that. Ideally, the lead software engineer should have 5+ years of experience.

Company’s past ML projects

If the company truly invests in Machine Learning, they must have solved a few problems with ML already. Ask them about those projects. It is completely okay if they have used off the shelf solutions or any third-party provider. Startups are tight on resources. For them, the quickest solution is more important than a novel one.

However, if they haven’t solved any good Machine Learning problems, it usually means one of these -

They don’t have the expertise
You’d be the first Data Scientist there
They don’t believe in ML and are just riding on the AI hype train.

Fin.

I don’t have anything against startups. I have worked at three. It is just harder to evaluate them. Especially if you are looking for your first full-time role.

How to avoid getting a data science job