Marco Caccin, Machine Learning Engineer at Stylight: this is what I do

Marco Caccin, Machine Learning Engineer at Stylight: this is what I do

We interviewed Marco Caccin, a Machine Learning Engineer at Stylight who explained us how he ended up in his role and plus many relevant insights.

Can you explain in simple words what your role as a Machine Learning Engineer is about?

To explain what I do, I first have to mention where I apply ML.

Stylight, my current company, is a search engine that aggregates fashion products from online shops around the world: each “thing” that we show on the website to the user must be tagged (example: “red t-shirt with print for men”) so that a user can find it by navigating or searching for something like it on the website.

My work is to create and continuously improve pseudo-intelligent machines that are able to understand what an object is given some raw information about it, like an image or a text description.

What led you to become a Machine Learning Engineer?

I kind of slowly but actively turned into one. From one point of view, my whole Ph.D. (in Computational Physics) was about making use of ML to make better and faster algorithms for atomistic simulations.

As I was progressing with my research, I realised that there were many more open questions and avenues for research that interest me in the field of Machine Learning rather than in the one of Computational Physics.

Another big part of the reason is that what really interests me is the understanding of consciousness, and I was naïve enough to think that by building Artificial Intelligence I would be able to explore and explain my own.

What do you like the most from your job? What do you like the least? What are the main challenges?

I really like it. My work is essentially Research & Development, so a mix of playing in the abstract realm of ideas and implementing stuff to see if the ideas make sense. Sometimes they do, and that’s when you feel useful because you are providing something of value to your colleagues.

What I like the least and I find the most challenging is about my field in general rather than my specific job: ML is a powerful tool and is very quickly becoming ubiquitous in our lives, but there’s rarely a thought about the ethical implications of deploying ML in a real world application. Take the example of targeted advertising to sway public opinion and (allegedly) change election results, or facial recognition for tracking citizens on the streets.

What’s hard in my job is making sure that what you are doing is OK for society.

Which advice would you give to any Machine Learning Engineer that wants to distinguish?

Everybody loves a person that is good both in Machine Learning and Software Engineering. If you are both, then your ideas will make it to the real world with much lower barriers, be it because you implemented your intelligent black box from head to tail, or – more realistically – because the software and data engineers that are working with you won’t hate you when implementing your code.

Which kind of skills would you need to train to make a difference in your profession?

Your biggest asset is domain knowledge: understanding the data and the business problems. Perhaps disappointingly, being able to create a very fancy machine learning algorithm is somewhat less important.

To understand the data you simply need to spend a lot of time working with it, no shortcuts. Also, be sure that you know who to ask when trying to understand what the data means.

Understanding business problem is again something that comes with time, as long as you keep your ears wide open 🙂

How do you think your role will become in the future? Which skills will you need? How do you stay abreast?

Perhaps I will lose my job in favour of automation because of an algorithm that I created, who knows…The truth is that it is really hard to make any prediction because this is a very new field (did my job position even exist 10 years ago?).

It certainly does not hurt to try and stay up to date with the new stuff.

Being a polymath will probably help too, so that your skillset is not limited to just one narrow field.

Are you looking for job vacancies as a Machine Learning Engineer?

Find out open positions on our website

2 Replies to “Marco Caccin, Machine Learning Engineer at Stylight: this is what I do”

  1. Hey there, I thіnmk your site might be hacing browser compatibility issues.

    Whеn I look at your website in Ie, іt looks fine but when opening in Internet Εxplorer, it has some overlapping.
    I ϳust wanted to give you a quick heads ᥙp!
    Other then tһat, wonderful bⅼog!

Leave a Reply

Your email address will not be published.