Before I begin, I think I should provide a little background information about my motivation for writing this article. I recently graduated with a Bachelor's degree in Computer Science. Halfway through my third year, I realized that AI was the only field I wanted to work in. My university didn't have any AI-specific courses, and there weren't many AI internships in Dublin. I am currently happy to be a graduate working in the AI R&D team.
In this article, I will briefly explain what I did to prepare for an AI job with no experience. Admittedly, none of this information is surprising and most of these tips are pretty obvious. Just like eating healthy and exercising, we all know this is necessary, but we find that many people still don't do it. We hope this article helps people create their own plans for moving into the exciting world of AI. Another thing to point out is that while this advice generally applies to all areas of software development, the specific guidelines and topics are clearly focused on AI/ML.
ML is a very unique field for software graduates and young developers. This field is still relatively young, only starting to develop in earnest in the last five years. This poses real problems/opportunities for both new developers and employers.
graduate: There is no solid information about what the field is like, there are few university modules and it is difficult to get relevant experience.
employer: It's very difficult to find people with relevant experience.
This is a difficult dilemma to overcome as a college student, but it is also a great opportunity. There is currently a significant shortage of qualified machine learning (ML) developers. Every company wants these roles, but they can't fill them. If you can demonstrate that you have relevant expertise, you will be a highly attractive candidate who will stand out not only to graduates but also to more experienced employers.
This sounds great in theory, but of course it's not that simple. There's a reason people like this are hard to find. This is a difficult field to master, this field is growing rapidly, and we have to catch up every month. Below, we'll discuss the key areas you need to address to build your brand as a machine learning expert.
- experience
- understanding theory
- Specialized field
- college
But the title says inexperienced? ? Oh my God!
I know that, but let's be honest: No one would hire someone with absolutely no experience.
As I said before, it's very difficult to gain experience working in a company that does ML. Therefore, if you can't do that, you need to get the necessary experience. Here are some good ways to do this:
- personal project
- hackathon
- coding challenges
- open source project
personal project
You must have 100% of your ML project on GitHub. This is a very easy way to remove people from the hiring process and will be the first thing a recruiter sees after your resume. Coming up with a project while you're still learning ML can be a little overwhelming, but that's okay. It doesn't have to be big, flashy, or innovative. It's enough to show people that you understand the topic and can work/research independently using appropriate coding standards. There are several things to keep in mind when building a GitHub project.
- Your project will not take more than a month to complete
- Make sure your code is clean, modular, and commented
- Provides a Read Me and other documentation about the code, including technologies used, tutorials referenced, dependencies, and more.
- If possible, provide unit tests for major parts of your codebase.
Choose the next project to work on. It should be simple enough that you can complete it within a month, but relevant enough that you'll learn useful skills along the way. Here are some examples.
- Image classifier using CNN. All you need to do is differentiate between two types of images (i.e. dog/cat).
- Standard feedforward neural networks for classifying data. Kaggle.com has a lot of great datasets. It takes a data set of iris, etc., and classifies the type of flower iris based on the given data.
- Sentiment analysis of movie reviews. Another popular first project is using neural networks to classify the sentiment (good or bad) of movie reviews. You can use the IMDB dataset, also available on kaggle.
hackathon
Hackathons are great for several reasons. You have to go out and build something, you can meet more experienced people, and you can put it on your growing resume and portfolio. In addition to looking for AI-specific hackathons, try participating in general software hackathons to incorporate AI into your projects. Check out meets.com to see if there are meetup groups in your area focused on AI or software development in general. These groups usually hold some kind of hackathon at least once a year.
coding challenges
Similar to hackathons, coding challenges require you to build a practical application of what you've learned. This is extremely valuable when applying for ML jobs. As an added bonus, these competitions are generally a lot of fun, and the added sense of competition is a great motivator. Take a look at Kaggle, CodinGame, Halite.io, etc.
open source project
This is the closest thing to real world experience aside from actually getting a job as an ML developer. Open source projects give you real insight into production-level code, and you can learn valuable skills like debugging, version control, developing with other people, and of course a lot of ML (depending on the project). can.
OK, the main thing is to get some projects and experience, but not just follow some tutorials and paste it into GitHub (please don't do that 😀 ) what you build You need to understand what you are doing. As many college students know, there is a big difference between studying something and understanding it.
There are many great resources out there that clearly explain the key theories of ML and deep learning. Another important thing is not to focus only on deep learning. I know that's the “sexy” aspect of AI, but it's only one aspect. Get familiar with more traditional forms of ML, such as regression models, support vector machines, and all the key concepts of probability and statistics. These are always valuable no matter what type of AI you're building.
Here are some of the better resources I've found.
- Stanford Machine Learning is a free course on coursera that covers just about everything you need to know about ML, from regression models to deep learning.
- DeepLearning.ai by Andrew Ng Like the previous course, Ng provides a very comprehensive introduction to AI, but this course is specific to deep learning rather than ML in general.
- Grokking Deep Learning by Andrew Trask is probably the best book on deep learning that I've found. This book explains how to build neural networks without using any libraries other than numpy. This is more complex and expensive, but can be very worthwhile if you take the time to think it through properly.
- Siraj Raval's YouTube channel is a great place to get an overview of almost any relevant ML topic, and it's also a lot of fun to watch.
This is a good way for employers to cut through the chaff. ML is a large field. It is impossible for one person to know everything. That's why people specialize and become experts. If you can show that you have a solid understanding of ML/Deep Learning in general and specialize in one area, you will be even more valuable to potential employers. However, be careful not to box yourself in. You don't have to decide on your entire career path now. Instead, find an area that you're really excited about, learn more about it, and if possible, try to do some projects in that area. . Below are some examples of specialization.
- Computer vision: CNN, segmentation, labeling, description, object detection
- Recurrent network: Time series data such as stock market and videos, LSTM cells
- Reinforcement learning: Teach agents skills like video games and driving
- Natural language processing: Chatbots, sentiment analysis, content generation, content summarization
- Generative adversarial network: Learn how to generate content such as images, 3D models, learning policies, and audio.
- Meta-learning: learn to learn
- One-shot learning: Learn with very little data
- Visualize and debug neural networks: Neural networks, a vast field of research, remain black boxes, and it's difficult to visualize them and understand why they don't work when they're broken.
This is a point that many people disagree with. The path to becoming a software developer is changing. College is no longer the strict requirement to enter the industry as it once was. Tech giants like Google and Apple have also begun waiving bachelor's degree requirements. This is because innovative companies recognize that the people they want to work with are passionate, self-motivated, and eager to take initiative. All of this doesn't explicitly require a degree, and with the vast amount of resources available on the internet, you can learn most of them on your own at home if you put in the effort.
That being said, getting good grades in college is a huge bonus and should not be underestimated. So my advice is that if your situation doesn't allow you to attend university, don't dwell on it, there are many other avenues open to you. If you're a college student, you need to crush it.
If you do everything else mentioned in this article, you have a good chance of getting a job even if you didn't do well in college. However, you will be in a better position if you can get a 1.1 (these are the highest grades according to the Irish education system) or a high GPA and achieve an excellent final year project (FYP) centered around one of the aforementioned topics. You will be able to stand on it. . Therefore, you should focus on this. Go to college with an aggressive mindset of “I'm going to crush this.” Attend all classes, take notes, complete assignments early, study hard, and get good grades.
This is the perfect time for the AI industry. Much like the advent of the internet, AI will impact all businesses regardless of domain and become one of the most popular tools for any organization. Currently, we are still in the early stages of understanding the potential of AI. This means the industry is in a state of innovation, discovery, and great uncertainty. There are very few experts and no one has all the answers. The AI community is constantly learning and improving. Therefore, if you take the time to learn as much as possible, you will quickly see how quickly you can improve your skills. It is not an easy task and takes time. So be patient, persistent and stay focused.
