 It's no lie that being a data scientist is probably one of the coolest jobs out there at the moment, particularly with the rise of AI last year. In this video, I want to detail my journey into how I became a data scientist and offer advice for how you become one in 2024. It's important to mention that not everyone's journey is the same, but hopefully this video will strike some inspiration or some general guidance on how you can become a data scientist. We'll start with covering my background, how I discovered data science, how I learned data science, how I got my first job and general advice for you looking to break into the field. So let's get into it. I come from quite a heavy maths background. My mum has a maths degree, both my grandparents studied physics and my great granddad was even engineer. So I was always naturally inclined and pushed into that direction of STEM subjects. I remember when I was 12 years old, it's already got me into physics. I was watching the Big Bang Theory and they were talking about things such as quantum blue gravity, general relativity. And these topics were really interesting, so I went and googled all about them. Obviously, I didn't know much of anything at that point, but they were still very interesting to me and that's when I decided to pursue physics. For my GCSEs, I got 4A stars, 3As and 5Bs, which was not bad at all. However, it wasn't exactly genius level. And at that time, I thought I was much smarter than I was and my work ethic was really poor. My 4A stars were in maths, physics, further maths, and chemistry, so those were the four subjects I took to A level. At A level, nothing much changed. I was still being quite lazy and wasn't working too hard because in reality, I thought I was smarter than I was. I was quite delusional at the time and I thought I could get into the best universities, even though, like I said, I wasn't working very hard. So in the UK, UK had a choice of five universities that you can apply to. So my five choices were Oxford, Imperial, Nottingham, Manchester, and Southampton. Now obviously, both Oxford and Imperial both rejected me as they are the top institutions in the world. At that point in time, my grades and work ethic, like I said, went very good. However, Manchester, Southampton, and Nottingham all gave me offers. I firmed choice Manchester, which required me to get A star, A star, A. And my insurance choice was Southampton where I needed 3As. So results date comes around for my A level results and let's just say I didn't do too well as I was expecting. I got A star in maths and being further maths by C in physics. So for someone who wants to study physics at university, that's not great. And Manchester and Southampton both rejected me. In the UK, we have something called clearing. Now clearing is where on results day, universities may have some space on certain courses and people who didn't get into their firm or insurance choice can apply to those universities. And luckily, the University of Surrey offered me a place to study physics for astronomy come September 2017. At Surrey's, where I really learned that there is no substitute for hard work. I know it's a cliche, but it's true. In my first two years, I had a much better work ethic than I did in my school days. And in my first year, I got a first and in my second year, I also got a first. At the end of my second year, I got accepted onto the master's programme. As part of the master's programme, you have to do a year of research. It's kind of like a mini PhD. And my year was done at the National Physical Laboratory in Teddington, or MPL for short. And my thesis was on measuring air temperature gradients using acoustic thermometer. During this year, I enjoyed it. However, physics research from that small snippet I had wasn't quite I envisioned it being like. And in reality, I found things to move a lot slower than I would have liked. And that wasn't quite for me. And I kind of fell out of love of physics. I still remember to this day exactly how I discovered data science. After coming back from a day of work at MPL, a video appeared on my YouTube homepage, and it was DeepMind's AlphaGo documentary on where they trained an AI bot to be the Go World Champion at least at all. After watching that documentary, I became fascinated about how they train this AI, like what algorithms did they use and what kind of process did they take. I was looking into reinforcement learning, deep learning, Markov chains, all these things. Obviously, at that point in time, I didn't understand everything, but I found it also interesting. So I looked online of basically opportunities and what kind of people or what kind of professions use machine learning. And that's how I stumbled across data science. Like most people, I had the age old question, how do I learn data science? Data science cuts and intersects into so many fields, maths, statistics, computer science, that it seems overwhelming. However, if you break down your learning to small chunks, it's very doable. Coming from a physics background, I pretty much had all the prerequisite knowledge I needed. I knew linear algebra, I knew calculus, and I knew statistics. So that means I could jump straight into the machine learning and understanding how the algorithms work. The first course I took was Andrew Nyg's course called the Machine Learning Specialization. I took this course back in 2020, and this is when it was still the 2012 version, and all the exercises were an octave or mat lab. It's been revamped and it's got more cutting edge algorithms in there, such as reinforcement learning, recommended systems, and it's also taught in Python. At this point in time, I only had experience in one programming language, and that was FORTRAN. So we got taught FORTRAN in my first two years of university, and for those of you who don't know what FORTRAN is, it's probably one of the oldest high-level programming languages out there. It was written in the 1950s. With it being my first programming language, it made me not really like coding that much, because everything was manual, hard, there's not many packages available for FORTRAN. Reflecting on it, learning FORTRAN was kind of a blessing in disguise, because it really got me to really think programmatically, and like I said, everything had to be done from scratch, and so when I went about learning Python, it was so much easier for me. The way I learned Python was by simply contacting one of the lecturers at my university who taught a computational physics course. Basically, I asked them for the course notes, and it was just an introduction to Python. In reality, any intro to Python course would have been sufficient. I also took the tutorial sprint Python course, and it basically told me all the things that was in those lecture notes. The main things I learned were Python syntax, functions, loops, classes, all kind of the regular things you need to know behind a programming language to design or build anything. I then went to learn a bit more of the data science specific packages, NumPy, Pandas, LaptopLib, and also Scikit-learn. These were done on the Kaggle courses, and these are very useful. These are kind of the main packages you would use day-to-day as a data scientist. After Python, it was then to learn the other language of data science, which is SQL. The way I learned SQL was that, again, I took the tutorial sprint online course to SQL. It took me around a few days, and to be honest, that course literally covers everything I use now in my day-to-day job. It teaches you all the basics and more some that you'll likely use in any interview and also in most jobs nowadays when you're a data scientist. After upskilling in machine learning, Python and SQL, I then basically started building some really simple projects. What I did is that I will get some data set from Kaggle, and I'll just randomly apply just loads of machine learning models to these datasets. I will link in the description below a lot of these projects, but comparing them to my abilities now, they weren't very good, but they allowed me to get my hands dirty and just try out a load of models. I built linear regression, logistic regression, decision trees, just a range of algorithms, and it really taught me how they work and how to apply them to a real life problem. The hardest part by far is securing the first job. You dedicate a lot of time to learning all these skills with the hope that you'll land that first role. I'm not joking when I said I applied to over 300 roles in my final year of university trying to get this first data science job. So when it comes to your first role, I honestly believe it's purely a numbers game. You really just have to put yourself out there, have practice in interviews, have practice in these takeover assessments to land that first role. I got my first role at an insurance company like on mid-level sites in the UK. It wasn't some fancy Fang or Quant Hedge Fund, like I said, it was just a regular firm that was really good and now worked with some amazing people. You don't need to work at one of these top companies, particularly at the start, because in reality, in some of these smaller companies, you may learn more because you may be asked to do more things, be more hands-on of a lot of the infrastructure. Like anything in life, it's really up to you to excel and put in effort. You can not grow at all in big companies, and you can grow a lot in small companies. The final thing I want to discuss is how you can stand out as a data scientist. In my opinion, these three things are very simple to do, and they give you so much more rewards than the effort you put into them. The first one is make sure you have a GitHub profile and populate it. On the screen is what mine looks like. Again, mine's not that fancy, but it does have some, you know, it looks good and it has some nice things added to it, what languages I know, what I do, and some basic repos on my past projects. You can do the same easily. In fact, just copy my template and add some basic repos of you basically learning Python. It does need to be too complicated, but I promise you, most people applying for entry-level jobs won't even have this. The second one is write a blog post. I'm still amazed at why people think this is so much harder than it really is. The goal of writing a blog post, particularly if you're just trying to land a job, you're not trying to make the post go viral. It's more just to showcase your learning and show that you're interested in your curious and willing to document your work. The simplest way you can write a blog post is, for example, say you learn how to implement functions in Python. Write a blog post about how you implement functions in Python. It really is how simple. Don't ever complicate it. The final one, which is a bit more tricky, and that is enter a Kaggle competition and do reasonably well. Now, what Kaggle will show to the employer is that you're able to break down a business problem into code in a data science way, and that's really useful because your job as a data scientist is to unify business with data and to solve that problem. The thing I want to stress is that there is no one best way to become a data scientist, but my journey hopefully gives you some inspiration or some guidance or even some tips on how you can tailor your learning or the steps you can take to become one in 2024. I highly recommend you action on the three key things I mentioned at the end of the video, that is, get a profile, write a simple blog post, and maybe we'll enter a Kaggle competition. These three things will set you apart from pretty much every other candidate, particularly for entry level jobs. So I really, really recommend you try them. If you enjoyed this video and want to learn more about data science and how to break into data science, then make sure you click the like and subscribe button, and I'll see you in the next video.