Q: Joaquin, you were enrolled in the first year of MIDS. That must have been exciting but also a bit scary since you didn’t know what to expect. Why did you choose Duke MIDS?
A: I chose the MIDS program for three reasons. The first one was the syllabus and the interdisciplinary perspective of the master. I felt that it was a good combination of core/fundamental courses and electives, something very different from other programs focused principally on theoretical aspects of data science. The possibility of doing an internship with real-life cases was the second reason that made me choose MIDS. It’s not only a great opportunity to get a post-graduating job position (not in my case given the fact that I was not allowed to stay in the USA after finishing my master’s due to my scholarship), but it was also a great opportunity to learn and improve as a data scientist. The third one was the value of the Duke brand. Maybe that one sounds a little shallow, but when you are an International student, that prestigious is a game changer.
Q: You graduated college in 2014 and took a few years to work. What experiences did you have during those years and what prompted you to go back for a graduate degree, particularly in data science? Would you encourage students to go straight to grad school after undergrad or to take some time to learn about their passions and interests?
A: When I graduated, I felt sure I wanted to pursue a teaching and research career in Neuroscience. However, during my first year as a research assistant, in which I was preparing to apply for a Ph.D., I began to notice that the activities that interested me the most involved areas related to data science (programming experiments, cleaning the data, performing statistical comparisons, etc.). Although I hesitated to change careers at the beginning, because I already had a plan in my head, doing so was the best decision I could have made. In my opinion, it is good to take some time to get to know yourself after graduation before making a decision. There’s no need to rush into such a big decision as enrolling in a graduate program right away. If we think about it from a data science perspective, taking some time to gather data allow us to make better decisions.
Q: Can you tell us a little about your internship and a bit more about securing an internship as an international student? What advice do you have for other students? (how to search for an internship, how to get the most experience out of an internship, etc.)
A: I’m not going to lie, getting an internship as an international student can be a stressful situation. It’s going to sound a bit cliche, but the best thing one can do is network. Applying massively to every internship available on LinkedIn is usually not the best action in terms of cost-benefit (at least not in terms of psychological costs for you). If I had to reapply for an internship, I would define a series of areas or companies that are of interest to me. Then I would try to connect with people who work in those areas or companies. These people will have more information about possible internships, or in the worst case, they will have information on where or who to ask.
Q: The Capstone Project is the first chance MIDS students get to work with partners on real data problems and essentially get a feel for what it’s like to work in the data science field. Can you talk a bit about your capstone?
A: My capstone was about inferring emotional valence using wearables fitness trackers and psychometric tests. It was challenging because it involved doing:
- Inter subject forecasting
- Intra subject forecasting
- Using demographic and psychometric data in combination with temporal data.
- Research of the state of the art.
I have a background in Psychology and Neuroscience, and this project was a good combination of traditional research and data science.
Q: You currently work as a data scientist at R/GA in Argentina. Can you talk a little about the projects you are working on? What programs do you use most? Are you autonomous or work in a team? Are you working remotely? What are the pros and cons of onsite and remote work when it comes to data science?
A: I am currently working as a Senior Data Scientist in the “Marketing & Data Sciences” department at RGA. My work is a mixture of consulting and product development. Consulting is a world by itself, sometimes we are asked to solve typical Data Science problems such as forecasting consumer demand for retail companies or churn rate for service companies. In other cases, we need to use data science as a commodity or a marketing experience. For example, creating a chatbot that will interact with potential clients in a marketing campaign or creating a Twitter bot to analyze tweets using NLP and provide useful recommendations to users.
When I say ‘Product Development’ I mean that we develop Services and Platforms as Software (PaaS and SaaS). We are in charge of developing, serving, and maintaining data services for the company. I cannot say much about these services, unfortunately. My technical stack is mostly Python (for everything), Tableau or Google Data Studio (to deliver dashboards to clients), Docker & GitHub for production, and lots of Serverless services from cloud vendors (Cloud functions, Cloud Run, Big Query, and Pub/Sub are invaluable allies).
In the Data Science team (that belongs to the already mentioned Marketing & Data Sciences department) we are a group of 7 people. We are 1 Software engineering intern, 1 Business Analyst, 1 Data Engineer, 1 Software Engineering, and 3 Data scientists. Despite these being the “formal” titles, everybody ends up doing tasks from other fields. We used to create micro-teams (i.e. 1 DS and 1 BI) to tackle some problems in a fast way when parallelizing the job is possible. I consider that this is valuable, besides being the fastest way of working we have so far because it allows us to learn things from other fields.
We are currently working 100% remotely, but we have the option to attend the office whenever we want. During the last year, we realized that some in-person presence is important to generate a good and motivated team and that is why we voluntarily decided to fix a day in the week to attend and work together. Working 100% remotely could be extenuating. You end up getting tired of having conversations with people on a monitor, your attention span reduces, and you start mind wandering (or at least that happens to me, haha)
Q: What courses prepared you for your data science career? Are there courses/electives you wish you had taken?
A: All the mandatory courses had been useful for my career. Currently, I’m working mostly with NLP and doing lots of MLOPS in the cloud, therefore “Data Analysis at Scale in Cloud” and “Applied Natural Processing” were the electives that helped me build the required fundamentals to work in these areas. I wish to have taken “Intro to Deep Learning” in order to have a better background to understand all the new models (diffusion, transformers, etc.) that appeared.
Q: If you had the opportunity to do the MIDS program over again, what would you do differently (if anything)?
A: I would choose different electives, or try to create a more defined path. Defining a learning path could be difficult If you don’t know well what you want (like in my case when I did the program). Therefore, exploring different electives and trying to discover what type of “data person” you are, is a valid option, but I would do it differently if I could now. As an advice, you need to decide if you want to learn about all possible “data roles” (generalist, computer vision, NLP, or MLOps oriented) to have an idea of what they are about, or choose one beforehand and deep dive into that. Both options are valid.
Q: What were the top 3 best things about the MIDS program and the 3 hardest things about the program?
- Freedom – You have plenty of electives to choose from. That allows you to create your own learning path.
- Resources – Duke has plenty of them. Do you want a virtual machine for a side project? Just ask for it. Need credits to use cloud services? you got it.
- Professors and staff – Everybody here is incredibly attentive. They really care about students’ needs and want the best for them.
- Homework – The program is a real full-time experience. Prepare yourself to spend lots of time outside classes doing homework or studying.
- With great freedom comes great responsibility – Having so many electives could be overwhelming, you are going to want to enroll in everything but that’s not going to be possible. If you enroll in more than you can handle it’s going to be detrimental. Don’t let FOMO take over you.
- Being far from home – This is relative, but if you are an international student like me you could end up missing not only your friends or relatives but also living in a culture that you are familiar with.
Q: If you could give any advice to current MIDS students, what would it be?
A: Try to enroll in an activity that sparks your flame. It could be data science related (such as journal clubs, research meetings from a lab, etc.) or not (a sport, volunteer, etc.). It’s important to have a space that motivates you and that allows you to meet new and different people.