Vetting Your Data Science Guru

Let's Light Twitter on Fire!

“Having an opinion is like having a dick. It’s cool if you’ve got one, just don’t go waving it in front of people.” - Asmongold

Asmongold Pictured Above

I’m not going to name any names here, but if you go on Twitter, or Quora, let’s say there’s a lot of these people.

Note: This advice only applies to those that live in North America. If you saw my Data Science around the world post. Then you know there are cultural and societal differences all over the planet. It does extend to the data science industry as well.

For example: In North America, the culture is not that formal, you can call your boss by his first name, and be done with it. In India, the culture is far more formal, so you’ll either call them “Sir” or “Mam”.

Table of Contents:

1. Writing Style

2. Real World Experience

3. Passion vs Skill

4. Course Promoters

5. Model Building

Rule 1: Writing Style

Formatting

If your Data Science Guru claims he’s an AI leader, but his writing style, and formatting is atrocious. You can go ahead and assume he’s lying and full of shit.

If you are a competent you will be writing reports for Portfolio Managers, CEOs, and VPs.

These people have things going on, and are actually solving real world problems. If you give them a report that looks like it was unformatted verbal diarrhea:

Take shot every time it says “ML” or “Machine Learning”

That report is gonna get sent right back to you, and no one is paying attention or reading any of this. Anyone whose is competent in the real world knows how to write, because if he didn’t he’d have been let go a long time ago.

If you write like the above picture, please go take some improv theatre classes. These classes are a nice way to get you out of your shell, and become more socially aware. Higher ups generally do not want to work with people they think are weirdos.

Inverted Pyramid Style

CEOs, Executives, Portfolio Managers, etc… are all busy. When they ask a question, they expect a direct answer, they don’t want to hear your life story.

If you want to be a Quant, pay attention: When a Portfolio Manager asks you to summarize the model you built into a quick report. The FIRST thing he wants to know is the most important information. In this case, this would be the excess returns (alpha) generated by the model, and the risk associated with it.

• If the returns were garbage: no one cares about the dataset, or the techniques used. The returns were garbage, so this shit isn’t to prod anyways.

• If the returns were great: Now, we are talking. Now he wants to know where the data came from. How hard would it be to setup a daily script for prod, and any potential biases you detected.

If your Data Science Guru doesn’t put the most important information at the top. But, instead puts it at the bottom, he’s never had to write a real report that was read by another human being.

In other words, he was never anywhere near the “AI leader” he claims to be.

Rule 2: Real World Experience

If your guru does not have any real world experience, you can go ahead and discard any career advice they give you. Why?

Would you ask a homeless man advice on financial management? Would you ask a single girl advice on how to get married? Would you ask a junior dev how to get promoted at work? There you go.

Someone who has never worked in the real world as a professional has no idea what it takes to succeed. This is because they themselves have not even succeeded yet. An easy way to see if your guru has real world experience is to put their name on google, or LinkedIN, and see what comes up.

Most companies will put the name of their current data science team members on their website. This makes it easy to snoop around anyways especially if this guy is a public figure.

A Data Science Enthusiast does not work in the real world

If you found the actual answer I referenced above. You’ll realize it was nothing more than an attempt to sell another shitty data science course. (Like we don’t have enough of them yet…)

Rule 3: Passion vs Skill

If you are trying to hit on someone, which of the following do you do?

• A: Tell them you find them attractive, and would like to meetup with them?

• B: Fawn over them, talk about how amazing they are, talk about how attractive they are, talk about how anyone would be lucky to get them as their bf/gf, etc…

The correct answer is A, and the wrong answer is B. But why though?

It’s because Person A is self-assured, and they are not bothered by whether the answer is Yes, or No. Whereas Person B doesn't get that many chances. So, they start trying to oversell themselves, and shoot themselves in the foot. The same thing applies when you are in an interview. Show them you have what it takes to do the job they want, and you are not bothered by whether they say yes or no.

At the end of the day, this is nothing more than a simple job. Someone people fix plumbing pipes for a living, while others hit keys on their keyboard. It’s nothing special, you are doing some stuff for your employer to help them make some profit to justify your salary + bonus.
Here’s an example of another Guru:

Anyone who says follow your passion has never made serious cash.

When you are in an interview, they are not going to ask you questions to see how passionate you are. They are going to ask you questions to see your skillset.

Ignore those who start telling you to talk about how passionate you are about data science. This is because they’ve never been high up enough in the chain of command. If you start trying to oversell yourself, you’ll make the other person suspicious of you, and you’ll get dinged. So don’t do it.

Talk about your skills, and pass the technical questions they give you, and move on. If Data Scientists were paid $50k/yr, we all know these passionate people would immediately leave.

Note: It’s fine if you enjoy working as a data scientist. The problem arises when you think you can trick the interviewer by talking about how passionate you are. People didn’t go up high in the chain of command (especially in IT) by being idiots.

The passion stuff only works in the low end. I assume if you are reading this, you don’t want to be stuck there.

Rule 4: Course Promoters

If you are not paying for a product, then you yourself are the product. One of the ways many of these “free” data science blogs make money is via affiliate marketing.

There is nothing wrong with being an affiliate. That is a legitimate business model. For example if you need an accountant, your friend can refer you to his friend and collect a tiny referral fee.

But, the problem arises when the affiliates promote something that is garbage. Here's how these free blogs operate:

1. Write content that is Search Engine Optimized (SEO)

2. Send as much traffic as possible to this website until you have 10k free subscribers

3. Take offers from cheap bootleg coding bootcamps and recommend them

4. The student gets no value, so they come back, rinse and repeat for more crappy bootcamps for profits $$$.

Think about it for a minute. If they solved your problem, and got you your data science role, you'd no longer be reading them. So, why would they solve your problem in the first place?

Especially, since it's "free".

Here’s a nice snippet from someone I stumbled onto:

Reads like a sales pitch, doesn’t it?

WTF is 1stepGrow?

Cool Logo

Oh boy another Data Science BootCamp. Nice!
Is the course curriculum any good though?

This is like the most SEO optimized course curriculum I’ve seen in a while.

Excel for Data Science… I think we are done here.

Note: I don’t have a problem with affiliate marketing. I have a problem when you try to sell yourself as a knowledgeable leader, and you sell garbage products. Your advice is affecting people's lives, have some integrity.

Rule 5: Model Building

Anyone who says a data scientist/machine learning engineer needs to know how to build models. Or says any stupid stuff about how you need to know all the details about how models work, etc…

Is a retard, or he’s lying to you to sell you his data science course. As of today, the modelling aspect of our job has already been automated:

Model Building in 3 lines

1. Import the AutoSklearnClassifier

2. Point it to your training data

3. Predict with it

And… Just like that, the model building is done. If you don’t believe me, watch this vid:

Read this post on what the real world looks like.