The machine reading and AI is among the most popular topics of these days, especially within the Tech area. I am lucky enough to work and increase these specialists every day as a machine learning engineer!
In this article, I will go for my journey to become a machine learning engineer, to waste the light and advice on your way!
My domain
In one of my formerly, I wrote more about my journey from school to get my first scientific job. I recommend you check that article, but I will summarize the key timeline here.
Everyone's beautiful in my family learned some kind of stem. My best friend was an engineer, both my grandmothers learned physics, and my mother is a math teacher.
Therefore, my way was always jumping on me.
I have chosen to study for a university, after viewing a great teaching of 12 years; It's okay that everyone was very proud!
At school, I didn't mute in any way. I was actually bright, but I didn't use myself fully. I got good marks, but definitely not what I did fully.
I was arrogant and I thought I would do well with zero work.
I applied to the high universities such as Oxford and the Imperial College, but I gave my behavioral work, I thought I had the opportunity. On the day of the results, I ended up cleaning as I missed my offer. This may have been one of the most sad days in my life.
Cleaning in the UK is when universities provide locations for students in certain lessons when they have space. Especially for the students who do not have a university.
I was lucky enough to be given the opportunity to read Physics at the University of Surrey, and I continued to get the first Masters degree in Physics!
Nothing can take a hard workplace. It's a cliche cliche, but it's true!
My real plan was to make PhD and become a full-time researcher or a professor, but I was doing the study year, and I heard the research work was standing. Everything was slowly moved, and there seemed to be a lot of opportunity in the space.
During this time, Deepmind took out their Alphago – Movie Document to YouTube, which came from my home to home.
From the video, I began to understand how AI worked and learns about neural networks, the strengthening of learning, and deep reading. To be honest, until today I have no expert in these areas.
Naturally, I dig down and find out that data scientist uses ALGORITHMS AI and a study device to solve problems. Soon I wanted and started to apply for Data graduation graduates.
I spent many hours in codes, taking courses, and I work on projects. I applied to 300+ activities And finally, the first Date Science Science Moppeence SCieveted September 2021.
You can hear more about my journey from podcast.
Data Science trip
I started my work in the Insurance company, where I built various readers of the guardian, especially using the tree packages that increase the Catboost, XGBOost, GLMS models.
I created predicting models:
- Fraud – Is there anything deception is making a claim for profit.
- Exercised prices – What is the premium to give someone.
- The number of claims– How many claims does one will have.
- Average cost of claim – What is the number of common application that one will have.
I made about six models that start to change and the separation of a space. I've learned much here, especially in the math, as I worked very much with nutrients, so my math knowledge was very good.
However, because of the company's structure and setup, it was difficult for my models to improve the previous Stage of Poc, so I lost my Toolkit's side and understanding how companies used machine.
After year, my former employer reached me and asked if I wanted to apply in the Junior Scientist's role in the most important data for predicting time predictions and the performance problems. I really liked the company, and after a few discussions, I was given work!
I worked for the company for about 2.5 years, where I became experted in predictive problems and interruptions.
Improve multiple algorithms and send my models to produce using the AWS using the best software methods, such as unit tests, low-line, CD plays, and much more.
It is worth saying that I have learned a lot.
I have worked very well with the software engineer, so I took a lot of engineering information and continuing to study mechanical and figures on the side.
I've even received a promotion from Junior to midward.
Converting to MLE
In time, I realized that the actual amount of data science applies to make live decisions. There is a good share of Pau Aborta Jo
ML models inside JYTERBOOOOOKS have a $ 0 business value
No point in building a complicated and complicated model if it does not reveal results. Seeking that extra accuracy is 0.1% by tying multiple models usually is not worth it.
You are better in building something that can be answered, and that will bring real financial gain to the company.
Through this mind, I began thinking about the future science of data. In my head, there are two ways:
- Previews-> It is primarily to determine understanding of the fact that the business should do and that it should be considered to look at its operations.
- Engineering-> Models, algoriths decide, etc.) Bring the business value.
I feel analyzing data science and create POC models that will end over the next few years because, as we said above, they do not offer a visible amount of business.
That does not mean that we do not call them completely; You should think about it from theory of their return from investment. FIRST, the amount you collect should be more than your salary.
He wants to say that you have done “X produced by Y”, which is the two ways that allows you to do.
Engineering side was so sweet and sweet for me. I really enjoy putting codes and creating people's benefits, and that they can use them, naturally, where I went.
To move to the ML side. I can get help from a software engineer, but I would write all the production code, make my formation of the system, and then set up the process of independence.
And that is exactly what they did.
I basically I became a machine learning engineer. I was developing my algorithms and made them produce.
I also took NEETCOOD data structures and algoriths course to develop my computer science and began to blog about software engineering concepts.
Automatically, my current employer contacted me at this time and asked if I wanted to apply for the General ML role and the performance of their company!
Call it good luck, but clearly, the universe told me something. After several cycles discussed, I was given a passage, and now I am an engineer that is fully read!
Fortunately, the way to get out 'fell to me,' but I created my luck and writing about my reading. That is why I always tell people to show their work – you don't know what it can come from.
My advice
I want to share advanced pieces of advice that helped me converted from the machine learning engineer.
- Experience– Developer's Time Developer – a position of entry in my opinion. You need to be well engaged in data science, machine study, software engineering, etc. You don't need to be a specialist to all, but have good foundations on the board. That is why I recommend to be a few years of experience as a software engineer or data scientist and other study sites.
- Code of Manufacturing– When you appear in the data science, you should learn to write a good production code, well-assessed. You should know things like typing, write, unit tests, formatting, teasing and CD / CD. It is not very difficult, but it just needs a certain practice. I recommend to ask your current company to work with the software engineer to get this information, it worked for me!
- Cloud plans– Many companies these days transmit most of its buildings and programs to cloud clouds, and the machine for machine-reading is not the same. Therefore, it is better to practice these tools and understand how powerful the models will go live. I learned a lot from this, honesty, but there are lessons you can take.
- Command line– I'm very sure about you I know this, but all technical experts should be competent in the command line. You will use it more when you send a production code. I have a basic guide to find here.
- Data Buildings and Algorithms– Understanding basic algoriths in computer science is very useful in the MLE roles. Especially because you will be asked about the conversation. It is not too difficult to read the learning and learning of a machine; It just takes time. Any course will make a trick.
- Git & gitemic– Also, many technological experts should know git, but as MLE, it is important. How to use Squash.
- Specialize– Many mle roles I have seen requires professionalism in a particular location. I look at predicting time series, prediction, and General ML based on my previous experience. This helps you to be outstanding in the marketplace, and many companies are looking for professionals these days.
The key value here is that I actually have my software engineer skills. This is reasonable as I already had all statistics, statistics, and a machine learning experience from being a data scientist.
If I were a software engineer, the conversion could be back. That is why the machine learning is a major challenge, as required variety.
To summarize with some thoughts
I have a free news book, To complete the informationWhere I share them with tips and weekly tips as a data scientist to do. Also, when you sign up, you will find my Free data science continuesincluding A short PDF type of my Roadmap!
