Contains thoughts and opinions from MSc graduates
About a couple of weeks ago, a curious individual reached out to me after reading an article of mine. The main reason for contacting me was to get my opinion on if a master’s degree was enough to build a career in the AI industry, or should that master’s degree be supplemented with a PhD.
The short answer is No.
An MSc in Machine learning equips you with more than enough knowledge of the domain to contribute in most practical environments.
But — yes, there’s a but — the level to which your career advances can be defined by the type of degree you hold.
There’s tons of hype around machine learning, and large numbers of individuals are flocking to academic institutions to gain degrees within ML — myself included.
The importance of making the right decision as to what level of advanced qualification you want to pursue cannot be exaggerated enough.
To help make decisions a bit easier and provide some clarity, I will make a case for pursuing an MSc, and conclude with my personal take on the debate.
I’ll utilise three criteria as a guiding beacon.
For those who are interested in the satements from the MSc graduates, simply scroll down to the ‘Knowledge’ section.
A master’s in machine learning is one of the best methods to build both theoretical and practical knowledge. In the end, you get recognised advanced degree.
The path to undertake an MSc in Machine learning can be quite smooth, as long as you have a background undergrad study in topics such as mathematics, computer science, physics, mechanical or electrical engineering.
MSc in the UK generally takes a year full time and undertaking the qualification on a part-time basis can take from 2–4years. In the United States, the average time-length of an MSc seems to be roughly 2 years.
I found some great resources that included the links and information of US-based universities that have MSc ML programs. The first resource is written recently by Stacy Stanford in 2020, and the article lists 10 Universities, while the other resource, written by MNM in 2018 covers 20 Universities.
Several MSc programs in machine learning take on a generalistic approach in the teaching of the field to students. This means that MSc students are exposed to a variety of topics such as robotics, linguistics, natural language processing(NLP), computer vision, programming, software design, signal processing, speech recognition and more. This generalistic nature of the qualification can be advantageous within an ever-changing industry.
Let’s use our guiding beacons to further analyse a case for taking an MSc in machine learning.
Earnings and Career Potential
The generalistic nature of MSc programs equips individuals with the versatility that a lot of employers are currently seeking. The advantage of having a general overview of the ML field is that you are not limited to the entry job positions that are advertised.
After completing my MSc, I was able to apply for entry roles in positions such as Data Scientist, Computer Vision Engineer, Machine learning engineer, and NLP Engineer. The reason I had the requirement of most of the advertised position is due to the fact that my MSc program introduces the topics in enough detail.
For the role of Machine Learning Engineer, about 30% of the roles I searched on LinkedIn required applicants to have an MSc. I used Paysa to get more survey insight into an ML Engineer’s salary. From a survey of over 300 ML Engineering roles, 17% of the profiles had MSc degrees, 12% had BSc degrees, and 28% had PhD degrees. Taking an MSc will definitely open more doors.
Analysing a survey of profiles of over 15,000 Data Scientists we can observe that nearly half of the profiles had Bachelors qualifications, and 14% held Master’s qualification.
Some might make the assumption that from the stats, a lot of Data Scientist roles just require a BSc, and this is true.
Data Science is not as technically demanding when compared to other ML Engineering roles. Although there might be a high demand for ML practitioners, there are also a large number of graduates looking to get into entry positions.
The key is standing out from the crowd, and an MSc definitely puts you ahead of counterparts with bachelors qualifications.
Now onto earning potential.
The same survey mentioned above for both ML Engineering profiles and Data Scientist profiles places the average salaries for both roles at $94,510 and $99,558 respectively. The top earners for both roles were gaining well over $120,000.
To be a high earner, it’s safe to assume that experience is a deciding factor, and another crucial factor could be how advanced the qualification in machine learning an applicant holds.
It should be mentioned that this survey is focused on profiles primarily located in the United States.
Impact
There is a substantial positive business impact that is brought through a well-educated Machine learning Engineer or Data Scientist flourishing within a team, business or startup.
This impact could well determine the survivability and success of an organisation.
An MSc equips an individual with the skills required to create and make a crucial impact within an organisation. The skills I’m referring to are not technical but soft skills.
For example, an MSc provides you with an opportunity to cultivate presentation skills through the processes involved in the submission of the end of study dissertation.
In my experience, I had three different occasions to present my dissertation work. One included an exhibition attended by professors and industry experts, and another involved a final presentation to machine learning professors and lecturers.
On the topic of dissertations, another impact an MSc presents students with, is the unique opportunity to apply machine learning to specific industries or processes, through projects.
During my MSc, students collaborated with hospitals and diagnostics department to create computer vision system that accelerated the process of classification of cancerous cells and tumours.
I undertook a project that involved the development of a computer vision system that performed pose estimation on quadrupeds(four-legged animals). The further aim of the project was to develop a system to help improve the gait analysis procedures within Veterinaries.
Knowledge
For this section, I decided to contact two individuals who have graduated from their MSc studies in Machine learning and Computer vision. Rather than providing my thoughts on the amount of knowledge you gain from an MSc, I passed the task onto them.
Below is a quick background on our special guests.
- Kausthub Krishnamurthy is currently a Computer Vision & Robotics Engineer, based in Sydney, Australia.
- Ignacio Hernández Montilla is currently a Computer Vision Research Engineer, based in Madrid, Spain.
Here is the statement from Ignacio:
“A master’s degree teaches you the theory behind all the hype with machine learning. There are currently a lot of ‘machine learning engineers’ and ‘data scientist’, that have only taken one or two online courses.
These online courses only introduce you to the attractive part of machine learning. A master’s, on the other hand, presents the opportunity of teaching you the knowledge a Data Scientist or ML Engineer requires, in addition to crucial mathematical thinking.”
Here is the statement from Kausthub:
“In order to get to a point where I could make informed decisions (for research or for business application) I needed to develop my intuition further by building my knowledge of a vast variety of computer vision applications from the ground up.
Yes — this means wading into neck-deep mathematics which, to be candid, isn’t my first choice of after-work light reading — and committing to a Masters program was one way of motivating myself to work on all areas of computer vision that would take me to the next level.
By the end of the MSc program, I found myself enjoying the mathematical rigour, being able to appropriately understand and evaluate the choices made in the latest research — enabling me to make informed decisions about the applications which I had previously just applied because that’s what the tutorial told me to do.”
By now, you probably have an idea of some key benefits to taking an MSc in Machine Learning. To level the playing field, I will provide some disadvantages that I believe are only fair to point out.
- Academic funding is always a significant factor as to why individuals choose not to pursue further education. An MSc course is definitely a costly investment, for example, the price of the MSc Machine learning course I took in 2018–2019 was £10,300. For international students, it was over £20,000.
- The adjustment period could be longer for some. Individuals that decide to go back into education might find it difficult to adjust to the academic lifestyle that involves long hours of study. People that have worked for several years before returning to education will definitely find the change drastic. Below is an article detailing my personal experience in regards to adjusting to an academic lifestyle.
- A period of not earning is to be expected with undertaking a full-time MSc, this might not be the case if you are in a part-time program.
- A weird feeling of getting left behind. For those taking an MSc at a much older age, during social interactions where your friends and colleagues are complaining about work-life, and non-academic related topic, it might create a feeling of isolation.
If you have the time and funding to take on a year challenge to advance your machine learning skills and knowledge, then taking a master’s in machine learning related topics could be the right decision.
There are tons of benefits to be realised by an MSc program, especially when it comes to career progression or change. This takes a lot of hard work, and that’s something you have to be prepared to handle.
As for those who might not be able to make a full-time commitment to any academic program, plenty of universities offer MSc courses on a part-time basis and are usually completed within two to five years. This option can also be a cheaper alternative for those with limited funding.
Lastly, it should be mentioned that an MSc does not entirely guarantee career progression.
For those who don’t have the luxury of considering institutional education, there are several online courses released on platforms such as Coursera and Udacity.
These online courses might not be as in-depth as an actual degree. Still, they do provide an individual with the necessary information and practical knowledge in machine learning. And a bonus is that many of the online platforms offer recognised certifications upon completion of courses. Here are a few below.
Thanks for taking the time to read this article, and I hope it was beneficial.
For those looking for more articles to read, below are a few other articles I’ve written that you might find interesting.
- Support my writing by becoming a referred Medium member
- Subscribe to get notified when I publish articles
- Connect and reach me on LinkedIn
- Learn with me at O’Reilly