How to switch to Data Science — 3 tips and many resources to prepare
After becoming a data scientist in 2019, I have often been asked by my friends how to start their careers in Data Science. Here I…
3 tips and many resources to prepare
After becoming a data scientist, I have often been asked by my friends how to start their careers in Data Science. Here I summarise the top tips and best learning resources I have used to share with them.
Tip 1 — familiarise yourself with free resources
Surprisingly, there are many introductory hands-on resources to learn Data Science from scratch that do not require specific payments, e.g., via subscription. Here are my recommendations:
Kaggle Learn — designed as a preliminary resource to prepare for Kaggle competitions, now it covers a quick overview of many relevant topics.
Free courses from Udacity — apart of their paid Nanodegrees useful to tailor industry experience (see below), Udacity provides a large number of free courses.
Free datasets for machine learning projects— there are many publicly available datasets, for example, Kaggle datasets for the above-mentioned Kaggle platform, sklearn’s datasets available for Python’s scikit-learn library, or datasets from aggregated dataset links via Google Cloud Platform.
Optional, if you have noticed gaps in coding (for example, in Python), I would recommend using hands-on platforms that also have free versions, such as Codewars and Leetcode.
Tip 2 — use the most efficient paid resources
Apart of free resources, there is a number of paid (either pay-at-once or subscription-based) resources. Here are my recommendations:
Deeplearning.ai contains a number of relevant courses/course specialisations developed by Andrew Ng and his colleagues since 2017, and placed (together with many other Data Science-related courses) in the Coursera platform that requires a subscription payment to access available resources and their hands-on platform. Currently, I have completed 14 of Coursera specialisations (as well as a few standalone courses), namely Applied Data Science with Python, Machine Learning on Google Cloud, Deep Learning, Data Engineering, Big Data, and Machine Learning on GCP, Networking in Google Cloud, Architecting with Google Compute Engine, DeepLearning.AI TensorFlow Developer, TensorFlow: Data and Deployment, Machine Learning for Trading, From Data to Insights with Google Cloud, Advanced Machine Learning on Google Cloud, Natural Language Processing, AI for Medicine, and AWS Fundamentals specialisations, and I can strongly recommend most of them (especially those from deeplearning.ai).
Udacity, in addition to a number of free courses, contains a number of the so-called Nanodegree programs. Currently, I have completed 5 of them, namely Machine Learning Engineer, Data Scientist, Natural Language Processing, Data Streaming, and AI for Trading Nanodegrees, and I can strongly recommend all of them (especially the first two that contain Capstone projects, being close to real-world industry projects, so you can safely use them as a relevant experience in your resume).
Cloud vendor certificates: I have worked so far with top-3 of major cloud vendors (AWS, Microsoft Azure, and Google Cloud Platform), and have been certified with all of them, this includes AWS Certified Machine Learning, GCP Professional Data Engineer, GCP Professional Machine Learning Engineer, Azure Data Scientist Associate, and Azure Solutions Architect Expert certifications, and I can strongly recommend all of them, specifically for the vendor you are looking.
Tip 3 — prepare for interviews
The last but not the least step. No matter how good your skills are, usually several rounds of conversations and/or coding is required to receive a job offer from your favourite company. Here are my recommendations:
Widely use your network for your positions/companies of interest, for example, you can find a common person to recommend yourself to hiring managers (or even know the hiring managers in person!), or at least you can figure out more details about the position/company of your interest.
Prepare your resume in a way consistent with the job announcement, for example, use the same wording and emphasise the required skills, ideally with numbers.
Prepare well for the “standard” interview questions — here, I highly recommend using at least top-20 videos (sorted by views) from The Companies Expert channel on Youtube, or using recourses such as leetcode.com to prepare for company-specific coding interviews.
Stay calm and professional during and after the recruiting process. Often, it is just a number’s game due to several favourable applicants with comparable skills. In this situation, application to more positions would simply multiply your chances of being hired.
Q1: Should I concentrate on “free” vs “paid” resources?
Well, nothing in this world comes “for free”. Even if you do not pay with cash, you still “pay” with your time and energy that you can use elsewhere (for example, to earn money). Thus, the decision criteria are very individual.
Things I usually consider for a new educational resource:
potentially added value (including gained knowledge and certification value — notably, both of them deteriorate with increased experience as the learning curve shallows);
expected resources to spend (time, money, energy);
competing “projects” (other projects, hobbies, family, friends, etc.).
Q2: Which cloud vendor(s) should I prefer?
I have worked with top-3 of major cloud vendors (Amazon Web Services, Microsoft Azure, and Google Cloud Platform), and have valid certifications from all of them.
My impression is that the major vendor’s ecosystems are gradually converging with each other, with similar products implemented around all of them (for example, their common machine learning platforms, such as Amazon Sagemaker, Google’s Vertex AI, and Microsoft’s Azure Machine Learning Studio).
Here are the tips to consider when choosing the vendor to learn / certify:
choose a vendor of your favourite company (if any);
choose a vendor from your previous experience (if any).
Q3: How many different skills should I learn to be hired as a Data Scientist?
Many variables that might affect the answer — available job descriptions for the role you pretend, previous experience, presence and strength of other competitors, your desired compensation, etc.
In my experience, the level of your competitors is the most decisive factor in whether you are hired for a given position. This includes not only “hard skills” we have discussed before, but also “soft skills” (the more important, the higher the role you pretend), and (not always!) salary expectations. However, this is not easy to guess as the hiring process is stochastic (and is definitely not that transparent!), so one usually considers alternatives.
For example, if a job description is available, I would consider this as a primary reference. However, in many situations, there are no job descriptions at all (for example, in the case of a “hidden job market”), or they are simply too generic. In this situation, my recommendation is to look for alternative openings for similar positions from several different places, and to summarise their requirements (alternatively, you might look at my favourite level description in IT, though not specific for Data Science roles).
I hope these results can be useful for you. In case of questions/comments, do not hesitate to write in the comments below or reach me directly through LinkedIn or Twitter.
You can also subscribe to my new articles, or become a referred Medium member.