If I Lose My Machine Learning Job Today, Here is How I Will Get A Better One.

I’m a machine learning engineer at Amazon.

Before that, I spent 12 years in biology research. In 2020 summer, I quit my Ph.D. without a plan for the next step. At the end of that same year, I landed a job as a machine learning engineer.

If I woke up tomorrow morning and suddenly lost all my job experience. Here is how I would get a better one.

1. I will focus on what I have and play by the rule.

If they still ask for LeetCode, I will do LeetCode. If they ask for PyTorch, docker, AWS, then I will learn those. So, go bookmark 30 jobs that seem desirable for you, and make a spreadsheet to answer these questions:

1. What are the most common technologies they are asking for?
2. What kind of projects can I do to prove I am familiar with those technologies?
3. What are the most common soft skills they look for?
4. Can I tell a story from the past to demonstrate that skill? If not, can I get an opportunity at my current job to demonstrate that skill? If I don’t have a job, can I be part of an open-source or non-profit and work for free?

All these things do not need permission. You can go ahead and do it; that is how I did it. During those processes, there is no way you will have less expertise or less experience. All those will show in your interview.

Doing this job market research yourself serves two purposes:

1. Data gives you confidence in your decision-making. Your research certainly will be incomplete, but the fact you researched everything firsthand gives you the saliency of these facts, motivating you better. Think about this: the fact you are interested in reading this article isn’t because what I am sharing is the most accurate or complete, but because my sharing is real. If you are human, saliency serves better than statistics.
2. The process of collecting data by yourself can bring you insight that you weren’t looking for initially. These insights prompt you to ask questions and understand the game better. When researching for my first job, I saw different companies choose different cloud providers. So I started wondering what makes a company choose between AWS, GCP, or Azure. Then I learned about the market share, the service history, and the customer strategy of each service, which led to my interest in earning an AWS certificate eventually.

2. I will take each interview as a stepping stone.

Whether this opportunity is a job or just an interview, treat them as an opportunity to learn rather than an achievement to conquer. Thinking of each opportunity as a stepping stone for your next move helps you in several ways:

1. This long-term horizon keeps you calm when the stake seems high.
2. It helps you focus on the part you can control since you have more control over what you can learn than how the result went.
3. It helps you focus on the most transferable part of this effort, so your overall return compounds the most.
4. To find your next step, you are encouraged to understand the bigger context and ask deeper questions.

External validation will always be

1. lagging
2. noisy

Therefore, you have to define your internal standards. To define those standards, you likely have to suffer humiliation and failures. It sounds awful, but it’s not once you get used to not giving a fuck and only focusing on growing.

3. I will ignore the nonsense opportunities.

I remember how excited I was when I got my first interview which later turned out to be a scam. At least, that became clear, and I stopped right there. However, there are more sophisticated scams that hide them as opportunities. Here is one example:

• 2+ years of combined academic/industry experience with predictive modeling, machine learning, advanced analytics
• Solid understanding and experience with advanced statistics and modern machine learning predictive techniques such as GLMs, decision trees, forests, boosted ensembles, neural networks, deep learning, etc.
• Fluent in Python, R, Java, and Typescript.
• Familiar with Git and CI/CD processes.
• Strong skills in data processing using SQL, Hive, Impala, Spark, Hadoop, Kafka
• Familiar with AWS services and tools infrastructure as code such as Terraform
• Excellent communication skills
• Passion for extracting hidden insights and building machine learning systems that enhance business outcomes

It’s practically impossible for an individual to be competent in all these skills

If something seems unreasonable, either in a good way or bad way, consider bending the reality first, not you. If you have all the good intentions and did an impressive amount of work, and things still don’t work, then the bug is probably systematic. If it’s systematic, it means you are facing a scalable problem, and many people share the struggle as you do.

You have no idea how much anxiety and self-doubt I had when I just started out my journey as a machine learning engineer, but I managed to not being taken over by them and stick to my schedule of doing my LeetCode and mock interviews anyway.

If you want to know how I landed my first job in 4 months without a degree, here is my 3 step recipe.

Talk to you soon!