From apple grower to fine arts student, from software developer to machine learning PhD- Jose Quesada has done it all. Now, he’s established Data Science Retreat, a course to help people with his passion for growth and development to delve into the world of data science. We recently spoke to Jose about his remarkable story, the Data Science Retreat experience, and why so-called “soft skills” are often the making of future data scientists.
Table of Contents
Give us a brief introduction to you & your work at Data Science Retreat.
I love developing people. When I see the ‘delta’ on a person after hard work, it makes me feel good. Data Science is ‘the’ place for this ‘huge delta’ development: because the state of the art is changing rapidly, you are forced to teach yourself new things every week just to stay current. Fields like this tend to attract people who like pushing themselves.
I came from a rural background. My father grew apples, and would expect me to do the same. Instead, I studied psychology and fine arts. Then I did a PhD with lots of machine learning. In it I developed a software system to teach pilots how to land commercial aircrafts without the need of a senior instructor sitting next to them (which I didn’t patent; silly me).
What you can see is that I changed direction many times; I taught myself mostly all I know that is really useful. I think we live in self-taught paradise. But after a certain level of excellence, it’s hard to make progress. This is something most aspiring data scientists find. No matter how many MOOCs you do, there’s a barrier that very few people ever break.
This is why Data Science Retreat started. I think I know how to create an environment where you can go “faster than average self-taught speed” and break the barrier of excellence that most people encounter. I asked myself: “What does it need to exist for this to happen?”. My answers was: you need to have access to ‘chief data scientist’-level people, contributors to leading open source packages, etc, and they need to be invested in your progress. You need to be surrounded by other people seeking excellence, too. DSR is the kind of setup that I wish I had when I started. Two batches later, all I can say is that I’m very proud of the result, as is everyone involved.
Talk us through the course structure.
You can check the instructional part online in our curriculum. What you don’t see there is how we approach the ‘portfolio project’, where you do original work under mentors.
We start with finding a good question. This is a creative process, and a skill you will use often once you graduate. Not all questions are answerable with data and machine learning. Of those which are answerable, not all of them produce business value. Once you know you have a good question, finding the data that can be challenging. Or cleaning it. Or making sure it’s correct.
Next step, you find a good evaluation metric (‘How do you know when you’ve won?”), and start iterating with your predictive models. When to stop fine tuning parameters is also a key skill; you will hit diminishing returns eventually.
Once you have demonstrated you answered the question you started with, it’s time to present your results and make a convincing case in front of stakeholders. Here, your communication skills determine everything: your beautiful product may never get put in production if you don’t do this well. You’ve exercised your communication skills quite a bit already by settling on the question; ideally the company is receptive, and was sold on the value. Do they believe you have generated that value, now that you’re finished?
At all times, you could have asked different mentors. You got around 270hrs of instruction on state-of-the-art methods. But let’s be honest: anything can happen here. It’s stressful. You are at the helm managing your project. Often you find your data has nothing going on for it, your predictive models are not doing anything interesting, or you cannot answer the question. You are back at square one. And there’s a hard deadline where companies will sit and look at you with their undivided attention.
What differentiates Data Science Retreat from other courses for aspiring data scientists?
1. Our mentors are at the ‘chief data scientist’-level or contributors to leading open source packages. All our mentors teach, and they are invested in your success. There’s nowhere else in the world you can get this today.
2. We focus on the question as much as on the technical details of the solution. We provide training on technical communication; you will present often, and get one-on-one feedback from a communication expert.
3. We prepare our participants for leadership positions. That is, either being the lead data scientist, or the only one in the company. This is far harder than preparing someone to join an existing group of data scientists and solve problems picked by someone else.
Why did you choose Berlin as the HQ?
There are two hotspots in the EU: Berlin and London.
Berlin has been doing really well in the last five to ten years with regards to Internet tech startups. When you look at figures in terms of size, how many VC-backed companies there are, how much venture funding flows into those companies etc. As a result the tech scene is huge in Berlin, there’s an interesting meetup almost every day.
London is also very interesting. There’s definitely money floating around because of so many banks. But tech-wise, choices are more conservative. If you are a bank, losing information, even if it’s a single transaction, is a big no-no. You have to stick to tried-and-true technologies. Berlin companies can afford to pick riskier, newer technologies, because they often deal with consumer-level information, which is usually not as crucial. If Twitter loses a tweet, it is unlikely they will get sued, unlike a bank. I suspect Berlin is already ahead of London tech-wise, and with time this difference will only grow. This is a good thing for data science, because companies who can take risks will use data scientists sooner than conservative companies.
What do you consider to be the main differences between the data science scenes in the US & Europe?
There’s of course a lot more VC money in the US, and this makes it easier for companies who use data science to exist. There are more web-scale, B2C companies in the US. 40% of the data science jobs are in the valley according to LinkedIn. And the pay is higher over there. So what’s to like about the EU?
- Since there’s less VC money floating around, companies doing well in EU (and hiring) are more likely to have solid business models (and be resilient to big changes in the economy).
- EU companies are less prone to follow fads.
- EU companies tend to offer better working conditions, even if salaries may be lower. Retirement, health insurance etc are all well covered by law. You get a full month of vacation. If you are on a high tax bracket, at least you know your money is not used for say fueling the military industry.
- There’s something to be said about being early days for data science in EU. There are better opportunities for truly outstanding people. From what I hear, there are in the order of 100 people able to do a good job at lead/chief data scientist in the entire EU. If you are one of them, or can imagine to be one shortly, you are clearly in a privileged position.
Still, I worry about EU competitiveness mid-term. Some companies are too traditional, and have trouble integrating data scientists in their structures. But this is a topic for another day :)
What are the essential skills and traits a data scientist must possess?
There are three must have skills to just enter the data science space. You need to know some programming of some kind preferably R or Python, but really any programming language will do. The second thing is that you absolutely must know some statistics and machine learning. This shouldn’t be a superficial understanding of these data analysis techniques – any programmer can blindly implement a technique as a black box. You need to actually understand why a particular technique is suitable and what its limitations are. Finally, you need to know how to query databases.
Different data scientists will have different strong suits. Some will be very strong with data visualizations, some with databases and others with statistics but all data scientists need to have these basic skills to work in this space.
We do not run coding tests, because nowadays with sites like stack overflow, it’s easy to write almost anything without really understanding the details. We consider coding tests non-discriminative. We do like to see code samples on github for existing projects.
We invite the most promising applicants to an interview. There, we make sure we are a good match for each other. There are questions about creativity with data, communication, and raw machine learning knowledge. We want to see people who have put the effort to learn this stuff on their own. Many interviews end early.
You’re currently accepting applicants for your third class; what level of prior knowledge do your candidates typically have?
You only need to know at least one programming language well. Other than that, there are no real prerequisites. We have applications from people who are already data scientists, but feel they are stagnating at work. Initial skillsets are all over the place, which makes is challenging (and fun!) to prepare the teaching. As you can see, the curriculum is very varied, and no participant has had experience in more than one or two topic groups.
A big chunk of people applying have been in the industry for years, and/or have PhDs. But I’ve seen many people with no experience, who weren’t so strong on paper, but ended up doing incredibly well during DSR. The sheer willpower and raw intelligence of some participants has been inspiring. I’m happy my interview process detected these people and let them in! I wish more and more people applied even though they felt intimidated by DSR’s reputation; if you know you have it in you, and have a burning passion for data topics, by all means apply! We tell you whether you are accepted the same day of the interview. If you are on the fence, I’ll encourage you to go for it; this batch we are hosted by Zalando, which is a great place because they have around 40 data scientists working already (two DSR alumni!).
(image credit: See1,Dot 1, Teach1)