What’s the Machine Learning Engineering Job Like
As a relatively new position, the day in the life of a machine learning engineer or data scientist is still a bit fluid. Find out what is like from people working today at Airbnb, SurveyMonkey, and Instagram.
By Alex Palmer
The following is a compilation of the thoughts of three data engineers/data scientists on what the machine learning engineer job is really like. It’s an excerpt of Springboard’s guide to AI/machine learning jobs.
What’s the Job Like?
Chirag Mahapatra is a machine learning engineer at Airbnb, Shubhankar Jain is a machine learning engineer at SurveyMonkey, and Mansha Mahtani is a data scientist at Instagram. Springboard sat down with each of them to talk about the job, so you have a better idea of what to expect day-to-day.
Chirag from Airbnb told us that the traditional software engineering role at the company is meant for keeping the product alive. In the case of Airbnb, the software engineer is focused on building new features that customers can see, and then ensuring that the platform has adequate back-end infrastructure to keep the site alive. Shubhankar from SurveyMonkey echoed that sentiment: front-end and back-end engineers work on UX/UI and build systems that are responsible for logic and business logic.
On the other hand, a machine learning engineer focuses on the machine learning infrastructure and machine learning models. Most machine learning engineers work on three main components: the infrastructure at inference time, the infrastructure for training, and infrastructure for annotation and labeling. In his year and a half at Airbnb, Chirag hasn’t deployed anything that consumers got to see directly. Instead, his work is focused on building models that shape their experience in a different way.
To get a machine learning job at Airbnb, Chirag said, you have to prove that you’re curious and willing to learn. Especially if you’re a newcomer. Execution ability is important for the company, and they want to see some background in machine learning—like building machine learning systems in production (and not just modeling). Understanding of different experimentation frameworks also goes a long way.
Shubhankar from SurveyMonkey shared some general interviewing tips for machine learning positions. According to him and others in the field, showing a list of projects that you’ve done in the past is important. A lot of people have tackled the Netflix challenge and the Titanic project, which are some of the most common ones, but have you done anything that isn’t commonly performed?
He suggested doing projects in the fields you’re interested in. If you’re into sports, why not do some analytics by scraping publicly available data warehouses, and building an interesting project for that? That makes you stand out on a resume. What makes you stand during an interview is explaining the project end-to-end, and the value that came out of its result.
Both Shubhankar and Chirag emphasized the importance of knowing not only the technical side of machine learning, but also seeing how the models fit in the organization, knowing the trends in the industry, and being invested in its future. One of the coolest trends in the industry, they say, is people and organizations sharing their work, tools, and knowledge. Uber first shared Michelangelo—their machine learning platform—and then released Ludwig, a toolbox for training and testing machine learning models without writing code. Airbnb published a paper on applying deep learning to search. These show that the trend in the industry is sharing knowledge and helping everyone get better.
On a typical day at Airbnb, Chirag interacts a lot with the product and ops teams, figuring out what’s missing and what the burning issues are. He also works on determining how new changes might affect the end user. At SurveyMonkey, Shubhankar’s day is split into three buckets of responsibility. The first is developing the infrastructure to support ML within the organization (automating retraining of ML models, automating how data scientists create new models, etc.) The second is maintaining existing models that are currently in production (making sure that the models perform as expected, and their quality doesn’t degrade). The third is getting new models into production (working with software engineering, product, data science, and data engineering teams to make that happen).
Springboard also discussed what the data scientist role entails with Mansha from Instagram. She said data science guides business decisions by aggregating data from many different sources and having proof that those decisions are data-driven. She was interested in the career specifically because she could be a part of and influence the direction companies take.
Mansha said that her typical day is split into three parts. Gathering the data is the first part. The second is storytelling—figuring out how to tell an exciting story about the data. The third is communicating with the team and evangelizing her thoughts, which is the most important part. The value of a data scientist is tied to whether or not the decisions she evangelizes are implemented. Storytelling and communication are crucial in taking an idea about what should be done all the way through to implementing it within a product.
She would also add: “Given both professions are relatively new, there tends to be a little bit of fluidity on how you define what a machine learning engineer is and what a data scientist is. My experience has been that machine learning engineers tend to write production-level code. For example, if you were a machine learning engineer creating a product to give recommendations to the user, you’d be actually writing live code that would eventually reach your user. The data scientist would be probably part of that process—maybe helping the machine learning engineer determine what are the features that go into that model—but usually data scientists tend to be a little bit more ad hoc to drive a business decision as opposed to writing production-level code.”
Bio: Alex Palmer is an entrepreneur and self-taught full-stack developer who has written extensively about UX design, communication, and AI. He believes that people are more important than software, and that AI can help companies build bridges between them by automating and improving lives and business outcomes.
- A Step-by-Step Guide to Transitioning your Career to Data Science – Part 1
- Projects to Include in a Data Science Portfolio
- Compilation of Advice for New and Aspiring Data Scientists