As a specialist Machine Learning recruiter, I know all too well the variety of ML roles that require different levels of industry and academic experience from the candidate. I was intrigued to find out what professionals (within the field) thought about this range of requirements, and which route they thought was the best to go down: academia or industry experience.
After speaking with CTOs and Lead ML Engineers, it became evident that opinions are very 50/50 and so, there isn’t really a perfect way to get into Machine Learning. Whether you spent more time in education or you jumped straight into gaining industry experience, you can still be as successful after taking either path.
Academia
If you are more of an academic person, then going on to do a Postdoc could be the one for you. It will take you through the foundations of tech within ML but more importantly, it will help you identify which area of ML you want to go into. Just a few areas you can look into are theoretical research, applied research or engineering.
There are many areas within AI and doing a Master’s in a STEM subject would enable you to gain good theoretical and practical knowledge in multiple subsets of AI; you will develop a deeper understanding of the data and algorithms, putting you into a stronger position to know how you want to progress in your career.
Industry Experience
While the academic route suits some, it doesn’t suit everyone – it definitely did not suit me! Many people say “doing is learning” and it’s more than possible to get into the industry by teaching yourself all that a beginner needs to know. If you are an enthusiast, you could teach yourself the basics: learn Python, watch YouTube videos, read papers and of course practice makes perfect.
A top tip when getting into ML/AI is to have your up-to-date portfolio on hand, ready to show. Work on some of your own projects, compete in Kaggle competitions and have them on GitHub or a similar platform. Show off your work for everyone to see! A Master’s is of course useful but, if you can prove yourself by outlining some of your top projects and demonstrate a solid understanding of the code, then why wouldn’t a hiring manager want to speak with you?!
Here are some key frameworks and languages you should look into when starting out in this space:
- Python
- Tensorflow
- PyTorch
- AWS/Azure/GCP
- Docker
- Kubernetes
- Airflow/Kubeflow
Give yourself some wide-spread knowledge in the field to ensure you stand out. AI companies are looking for a wide range of skills in their employees and how you home in on those skills depends on what route you would want to go down: academia or industry experience.
If you are looking to be a successful engineer but haven’t been to university then of course, teach yourself all the appropriate things you would need to know. However, if you prefer the research side of things then maybe going to university and gaining a PhD would be your best bet.
Ultimately, you need to decide where you want to be in 5 years time, whether that’s researching for a top university or you’re at a senior level in a scale-up. Once you have decided this then you can choose whether academia or industry experience is the best direction for you. There is no right or wrong answer to this matter, it’s all down to you and what you think you would excel in doing.
----------------------------------------------
I would like to thank the following people for their input and help with this blog, you have all been a huge help with writing it and I hope this blog can help others decide the path they want to go down!
Leon Fedden – AI Deep Learning Lead at AstraZeneca
Thanassis Bantios – CTO at Beat
Robert Smith – Lead Data Scientist at Clarity AI
Josh Eadie – Co-Founder and CTO at Measurable Energy