Section 1.1: Job Prospects and Requirements

So, how feasible is it for you to become an ML researcher or engineer earning six figures at a leading firm like OpenAI, Google, or Meta? The blunt truth is: it’s quite challenging, though not unattainable.

Factors such as your experience, enthusiasm, networking, and, notably, luck, all play a role. A common question among aspiring ML researchers is whether a Ph.D. is necessary for obtaining a position at a top company. The straightforward answer is: not necessarily, but it can be very beneficial. Alternatively, relevant experience may suffice, which I can elaborate on if there’s interest.

While a Ph.D. is not mandatory for most ML engineering roles, it often serves as a prerequisite for research positions. This is logical, as research roles demand domain-specific expertise, typically acquired through years of dedicated study on a particular issue. A Ph.D. provides the invaluable opportunity to spend 3-5 years honing in on a specific research topic.

In the hierarchy of roles, a machine learning researcher, or research scientist, generally holds a higher position and earns more than an ML engineer. The latter typically assists in implementing the researcher's concepts, particularly in larger companies that differentiate between these roles.

However, possessing a Ph.D. doesn’t guarantee success. I know several Ph.D. holders whose skills vary significantly. Some are exceptional, while others are merely average or below par. Proving your worth through publications, collaborations, and clear communication is vital, and this can be a harsh reality for many.

For instance, I’ve encountered Ph.D. holders who are incredibly talented yet humble, enjoying their research journey without the pressures of being overly ambitious. They thrive in their roles at esteemed institutions like Google, driven by genuine passion for their work.

Ultimately, you must have a deep love for your field to sustain motivation over years without immediate financial reward, or you need to discover this passion early in your academic journey, showcasing outstanding work through personal projects, open-source contributions, or leading publications.

The path to becoming a successful ML researcher is not a quick one; it often requires several years of dedication, and a fair amount of luck plays into the mix.

Section 1.2: The Importance of Coding Skills

Let’s shift our focus to a crucial skill in ML research: coding. In earlier times, physics research often involved small teams generating theories with basic equipment. Fast forward to today, and to contribute meaningfully to modern physics, one must often work at a facility like CERN, equipped with advanced technology.

Similarly, ML research has evolved. A few years ago, developing ideas could be accomplished with relatively simple coding. A landmark development in deep learning, ResNet, exemplifies this; the code modification was minimal yet impactful.

However, with the rise of large models, researchers now need extensive knowledge in distributed computing, hardware optimization, and managing large-scale infrastructures. This is why many big firms delineate the roles of research scientists and ML engineers.

Fortunately, there are numerous ML domains that don’t necessitate working with massive models. Although discussing each would be extensive, I have a comprehensive list of these domains available for free.

Depending on your preferences, the depth of software knowledge required in ML research may excite you. However, if you’re inclined towards research to avoid engineering tasks, you may need to reconsider your focus.