Srdjan is a mentor with Springboard’s online AI/Machine Learning bootcamp and – as a self-learner himself – he loves teaching the next generation of technologists. We caught up with Srdjan to learn about how he broke into Machine Learning, why he enjoys being the “hype man” for online bootcamp students, and why he believes there will be no shortage of jobs in Data Science for the foreseeable future.
What is your background and how did you get into Machine Learning and AI?
I first got exposure to statistics and data in school – I have a bachelor’s degree in Economics and a master’s degree in Econometrics from the University of Belgrade. After graduation, I landed a statistician job with The Nielsen Company, the marketing research consultancy, which taught me everything I know about data analysis, data cleaning, exploring data, and building predictive models – now known as Machine Learning and Data Science.
The leadership team at Nielsen emphasized outside learning in both theoretical and methodological views. We got together with statisticians from the other offices in Central Europe for eight hours a day for six days per week – what would now be called a “bootcamp” – and covered a series on statistics or data mining with case studies and a take-home exam or mini capstone project that would go into your performance assessment. Those immersive classes were part of my motivation to mentor at Springboard.
Now, I have five years of experience being a data scientist, a total of twelve years of experience working with data, and three and a half years as a mentor for Springboard.
Why did you want to become a mentor with Springboard?
I’ve always loved teaching and I wanted to do online mentoring to contribute back to the community. Self-study is very hard – I started learning Python through a 12-week course from EDX and took additional courses on Machine Learning through EDX and Coursera. When I was doing it, I didn’t have any friends who were programmers so I didn’t have anyone to talk (or complain) to. I felt like a program that provides student mentorship would be a great way to give back to the data science self-study community. It’s been the most fulfilling and rewarding experience of my professional career – hands-down.
Do you also have a full-time job in addition to mentoring at Springboard?
I recently started a consulting company, Logikka, with a friend (also a Springboard mentor)! We provide end-to-end Data Science and Machine Learning solutions to clients in order to help transform their business, including in-house training at companies here in Serbia. As the co-founder and Principal Data Scientist, I lead and oversee the more technical parts of our projects (but being very hands-on).
As a mentor, how do you work with your mentees at Springboard?
Each student has one 30-minute call with their mentor per week at a designated time and we also stay in touch over email. I have access to the Springboard dashboard to monitor my mentees’ progress throughout the curriculum so I can see how they’re doing in the bootcamp and be prepared for the call.
Springboard students in the AI/Machine Learning Career Track have a lot of support – mentors can answer questions over email or in the community forum, Teaching Assistants (TAs) are available, and they have Unlimited Mentor Calls where students can reach any number of mentors. If you’re working on a homework assignment or mini project and are really stuck, you can easily contact a mentor for help.
How is mentoring for Springboard different from teaching?
Mentoring itself isn’t teaching – mentoring is guiding someone by lighting the way. Yes, we can get more hands-on if needed, but Springboard has other academic resources to help with that portion. There’s also a fine line between mentoring and tutoring – for example, in academia, a PhD student is going to meet with his or her advisor to understand something from an academic paper, not a specific math concept, expecting the advisor to teach them something using chalk on a whiteboard It’s about guiding students through the curriculum, the mini projects, and the capstone project, and making sure they’re on track to graduate and aren’t falling behind.
Sometimes, mentoring is also being an ear to listen when my mentees need to vent – six months in a bootcamp is a long time, so if they get discouraged, I’m their “hype man!”, and am there to help them recognize how much they’ve achieved so far. I’m also able to shed light on the Data Science career path since I follow the industry in its technical progression, how the Data Science role is evolving, and the latest job requirements in the market.
Do Springboard students work in addition to taking the bootcamp or is it a full-time commitment?
Springboard’s AI/Machine Learning bootcamp is primarily a self-study program and most of our career-tracked students are in-between jobs and are fully committed to the course. Some do have full-time jobs and families, but some might have a part-time job and the rest of their time is focused on Springboard. Most people will finish in 6 months if they spend 15-20 hours per week on the course.
What types of students take the Machine Learning/AI bootcamp?
There are two groups of students that take this bootcamp:
- First, experienced Software Engineers who want to get into machine learning because lots of software products (from web services, to phone apps, and even physical products that, of course, run on software) have a machine learning model at their core. If they’re going to be the one integrating these models, they want to know how to build them and add to their full-stack skills.
- The second group are Data Scientists who have been working for a few years but not building software and want to up their game because more employers are looking for full-stack data scientists – people who can build a model and also package it up properly so the engineering team can just “plug-and-play” the model.
Because your students have professional experience as a software engineer or data scientist, are they getting jobs as Senior Machine Learning Engineers when they graduate?
By its very nature, Machine Learning Engineer jobs are “senior.” After graduating, students will be qualified for a senior level engineering position in Machine Learning.
What goes into Springboard’s Machine Learning curriculum?
Each unit has mini projects (homework) and hands-on work. And of course, since this is a Career Track, there are a number of units on career-related topics like networking, looking for jobs, and writing resumes. The Machine Learning Bootcamp Curriculum includes:
- Software Engineering – a foundational unit so the Data Scientists can learn the concepts and, the Software Engineers can refresh them
- Data Engineering and Code Optimization - because you need to know how to bring in the data and store it, as well as write fast, optimized code when working with very large datasets
- Data Wrangling – all the different ways of acquiring, cleaning, and analyzing the data in various shapes and forms, including doing it at scale.
- Math and Probability used in Artificial Intelligence, covering the “engine” behind deep learning – linear algebra, calculus, and probability and Bayesian statistics.
- Foundations of Machine Learning – looks at Machine Learning algorithms with a more mathematical view, together with the software tools and Python libraries for building models
- Machine Learning at scale covers Spark, the primary Big Data engine in the market, and how to do data processing and Machine Learning on it.
- Deep Learning - the mathematics, applications and tools behind Deep Neural Networks, that have powered amazing advances in image classification, object detection, automatic language translation, creating a text message by dictating it into your phone, chatbots, and virtual assistants (like Siri and Amazon Alexa)
- Mandatory unit on text, analytics, and natural language processing, and a unit on computer vision and image recognition
Finally, there’s a unit on the ethical and philosophical aspects of AI and its rapidly growing omnipresence in our lives, plus a look ahead on what’s coming next. Machine learning is going to be everywhere – in our daily lives, our society, our governments, our financial institutions. It’s going to be in the technology we use, it’s going to impact our economy. We want our students to be literate about data and aware that decisions are being made by algorithms. If we train algorithms using historical datasets that are already biased, then we’re training bias into the algorithm, and merely automating existing social injustices - or, as Holden Karau once said: “we’ll just burn the world down faster, but with data”.
Could you give us an example of a Machine Learning capstone project?
Quora hosted a competition to build the best model for detecting similar/duplicate questions on their website. One of our students could reproduce something like that. It’s text data so there’s a lot of data wrangling and cleaning to be done. It allows you to use both the traditional NLP approach with established ML algorithms, but also use new Deep Neural Network approaches and compare the results. You can wrap it up into a RESTful API service that could pull questions out of Quora and say whether they’re duplicates or not in real time.
You work in data science – in your opinion, does the Springboard curriculum match the needs of the evolution of the job market?
My answer might be a bit biased because I helped develop the Springboard curriculum! Our advanced course teaches you both the standard concepts, as well as cutting edge technologies. The career track curriculum gets a major overhaul once per year – for example, I recently helped enhance the Data Science track and added a unit on Software Engineering best practices for Data Scientists). Data Scientists need to be able to work closely with Engineering teams, and now the Machine Learning Engineering track is an extension of that with cutting edge material.
What are some of the challenges your Springboard mentees have encountered?
Self-study is hard – you’re in front of the computer all day, trying to understand complex, new material. If you have an off day and a mini project is taking longer than anticipated, you can get in your own head and start doubting whether you’re good enough or whether the career is right for you. The Imposter Syndrome kicks in, especially with students who are making career changes and don’t come from traditional STEM backgrounds. We have students from the social sciences who didn’t have as much math or programming experience but became excellent Data Scientists because of the course’s strong methodological approach. I’ve been through the same things they have so I understand the emotions they’re going through – whether they’re able to verbalize it or not.
Why do you think Springboard is the best bootcamp option for Machine Learning?
The curriculum has been carefully curated to offer the best that there is out there, to teach cutting edge concepts, and has been not only curated by very experienced Data Scientists, but also by Instructional Designers who are experts in adult education. I think it’s the best option because while it has a six-month deadline, it still allows for flexibility from week-to-week to accommodate for life interruptions. I also think the sheer amount of support that you get from the Springboard team – mentors, TAs, mentor calls, student advisors, career coaches – and an active online forum where I have seen more activity in the machine learning career track with students asking questions and problems they’ve encountered, as well as showcasing their solutions afterwards.. Students answer students, TAs and mentors pipe in when needed – there’s a real sense of camaraderie that exists which is extremely important because of the inherent difficulty of the self-study course from a psychological and motivational point of view.
When I started mentoring, Springboard was the only option, but now, quite honestly, I feel like Springboard is THE best option for students interested in a Machine Learning bootcamp and for professionals interested in mentoring, especially with the job guarantee.
Springboard’s Machine Learning Career Track is for applicants who have experience in technology. What is your advice for beginners who are interested in a career in Machine Learning?
I would suggest brushing up on math, especially a calculus or statistics course at the first year college level. Having a college degree isn’t required to do the course but some larger employers do require a college degree for hiring. Additionally, practice your programming skills - the Springboard entrance exam consists of three non-trivial programming problems, so your “coding game” needs to be at a good level. If you don’t know anything about programming, you should take a programming course first and find some online resources to practice your skills. You can reach out to the Springboard admissions team for recommendations.
We have entered an age where there will be no shortage of jobs for people interested in, and are comfortable, working with data. The level, complexity of the work, and position will depend on your knowledge, required skillset, tools that you’re proficient in using, and the work you put in, but if you want to work with data, you will not be unemployed.