March 25, 2022
This is the second of a three-part series that spotlights software engineers participating in Meta AI’s Rotational AI Science and Engineering (RAISE) program, which aims to give participants from diverse backgrounds an opportunity to start their AI journey full-time at Meta. See the first part of the series here.
Over his 10-year career, which by 2021 had spanned three continents and some of the biggest companies in the tech industry, software engineer Mohamed Gamal had grown more and more interested in AI. It powered every product he had helped build, and he began to realize AI must hold the key to better solutions for even bigger problems.
Mohamed was working as part of Meta’s creator wellbeing team when he heard about our Rotational AI Science and Engineering (RAISE) program, designed for software engineers who want to kickstart a career in the field but lack professional AI experience. Over 18 months, participants — called RAISErs — rotate through three world-class research and engineering teams at Meta AI and gain hands-on experience with a variety of cutting-edge technologies. Joining the program meant a significant career pivot, but Mohamed knew the opportunity was too good to turn down. “Don’t fear the change,” he said. “Trust your gut and take a leap of faith.”
In round two of our three-part RAISE Q&A series, we check in with Mohamed Gamal and Fernando Hernandez, RAISErs from the same cohort. Fernando, based in Seattle, had significant experience with large-scale database systems before joining the program. Mohamed and Fernando share how they learned about RAISE, how they found themselves at Meta, what they’re currently working on, and where they see themselves after the program.
Check the first post of the series here.
Want to learn more about RAISE? RSVP here for the upcoming virtual panel and info session on March 31, 2022.
Applications for the 2022 RAISE program will close on April 15, 2022. For more information and to apply, visit our program page.
Mohamed Gamal: After I got my B.S. in computer science from Cairo University in Egypt, I relocated to the United States for a yearlong internship at Meta. From there I went to Google Shopping Ads in Zurich, where I learned the fundamentals of engineering best practices. After three years working on infrastructure and backend, I wanted to build features and products that I could see people use. I transferred to Google’s Mountain View office and stayed until mid 2020. After almost six years at Google, I moved to Meta, where I worked on the Facebook App Creators wellbeing team. It was an eye-opener to see the complexity involved in solving integrity problems, given the diversity of our users and the differences in policies from country to country.
I I joined RAISE in July 2021. The transition to AI was a really new experience.
Fernando Hernandez: I grew up and attended college in Mexico. My major was computer engineering, but late in the program I realized I’d find more opportunities writing code than designing chips. I went to Germany for my master’s in computer science, where I learned about traditional machine learning — deep learning wasn’t taught widely yet. My courses were highly theoretical, mostly math and experimentation. Even though machine learning was starting to take off, I didn’t enjoy the math-heavy aspect. I consider myself a builder, and I found it more rewarding to build systems that ship to production. AI seemed like more of a research activity.
After grad school, I moved to the United States to pursue traditional software engineering at a database startup. Then I worked on large-scale database systems at AWS — my first big company job. I learned a ton and was exposed to industry-leading software development practices.
When I was offered a position at Meta, I was also given the chance to join the RAISE program, where I could learn how the company builds large-scale AI systems. The opportunity came at the perfect time, just when I was looking for a challenge in a new field.
MG: AI has been a core component of each product I’ve worked on. I realized that experience in the field would inspire new ways of thinking, so I could solve more problems and build better products.
FH: When I was deciding whether to join the RAISE program, I talked to my prospective manager about what their team actually builds, and I finally understood that there’s more to AI than research. The idea of developing large-scale AI systems won me over.
MG: AI is a large field with different domains. The chance to rotate through Meta’s diverse teams — natural language processing, computer vision, ads, ranking, integrity, infra — is giving me a taste of the different aspects of AI, so I can decide which domain to concentrate on.
FH: Meta is a clear leader in this field. AI is part of the foundation of the products we ship so I was excited by the prospect of working on those projects, both during and after the program.
MG: I was new to both the team and the domain. I told myself beforehand that the program is a step in my career, so I had more tolerance for the learning curve and didn’t feel bad about it. Because the RAISErs are all new to AI, the first month of the program allowed for formal training and co-learning. That time at the beginning just for ramp-up, without the stress of managing anything else, allowed us some welcome breathing room.
“Like any company of this size and pace, there’s always more work to be done than time to do it in, but the encouragement and advice from my manager, peers, and collaborators has exceeded my expectations.
I’ve had great, dedicated mentorship. My manager put me on a project that stretched me just enough. My peers really want to help, and they’ve saved me a lot of time by summarizing their experiences.
FH: It’s been amazing! I’ve learned so much. Like any company of this size and pace, there’s always more work to be done than time to do it in, but the encouragement and advice from my manager, peers, and collaborators has exceeded my expectations. That strong support network helps me align my goals with my team’s expectations, so I can achieve more with less effort.
MG: I’m learning how to solve product problems with AI core and abstract tech. In the first rotation, I worked on natural language processing. I got hands-on experience with every stage in the AI development life cycle: data exploration, data collection, modeling, and production. Each step has been an education, from balancing supervised and synthetic data labels to evaluating the trained model. I’ve worked with different technologies and codebases over the course of my career, so I can ramp up fast on any code and quickly understand the logic flow of each part of the system. That has helped a lot.
In my second rotation, we’re using AI to place and style text captions on display ads, so I’m learning the fundamentals of computer vision as well.
“I got hands-on experience with every stage in the AI development life cycle: data exploration, data collection, modeling, and production.
FH: My focus has been learning about the AI platform and infrastructure stack at Meta. I’ve worked on teams supporting different stages of model development: one in inference and the other in training. As it turns out, there’s a large gap between developing models for research and actually deploying them at scale. Some of the skills I’ve learned from productionizing other kinds of software are in huge demand in this kind of work, which has made my ramp-up somewhat easier.
MG: Either staying within applied AI or something in product machine learning. Applied AI aligns very well with my interest in building products.
FH: I hope to help solve hard AI problems. I’m at my best building infrastructure, but designing with other users in mind would certainly strengthen my skill set.
MG: It’s the sweetest spot when the team’s goals match yours, so pick the team that’s doing what you want to learn. But even if a particular rotation turns out not to be the right field for you, you get a new experience.
Also, don’t fall in the trap of thinking you won’t be able to contribute to the team. Offer them your expertise — in coding practices, leadership and ownership, people skills, you name it.
FH: Ask a lot of questions about how the program relates to your experience and goals. It’s not all research — but it can be if that’s your thing. As I discovered, there are many paths for people from different backgrounds to break into AI.