Born and raised in Paris, I come from Persian origins. As a kid, both my twin brother Afshine and I loved math and science, and we both wanted to be commercial airline pilots. I loved the idea of flying, and I liked the pilot mindset – very calm, composed, and logical. I was still dreaming of flying when I entered engineering prep school after high school. When it came time to sit for exams for university, however, there was a downturn in the number of pilots being hired, and I really wanted to explore math and science further. That all led me to study engineering at École Centrale Paris.
During that time, I had an opportunity to do research in a computer vision lab, which opened the door to discovering new deep learning techniques and made me believe Stanford’s ICME program aligned perfectly with my research interests. I was attracted to ICME’s intersection of math and computer science – to the idea of learning how to make things work from a mathematical perspective, as well as being able to execute those programs computationally. The combination of both these fields is what makes this program very powerful.
A highlight for me at ICME was being able to work as a TA for AI-related classes. My brother and I wrote study guides, which are still being used by students today. I found that helping communicate these topics to others was exciting, and made me feel I was making a positive impact.
Before completing my graduate program, I worked as a data science intern at both SLB and Uber. I joined Google in 2019, where I’m now a senior software engineer. First I worked as part of a team focused on Google Assistant, evaluating the responses the system generated and using machine learning techniques to find the best response to return to the user. Now I work as part of the Gemini team in Google DeepMind, which powers these experiences using large language models. Since 2021, my brother and I also have been teaching an ICME workshop on transformers, which is the same technology that powers the AI technology I work on. During the current spring quarter, we have been offering CME 295– Transformers & Large Language Models at Stanford.
Working in this industry has its challenges and rewards. Because the field is competitive and fast-moving, people work harder and longer than industry standards. It’s important to meet the moment, given customers’ excitement about the latest products. But it’s very exciting, and the AI technology we’re working on enables hundreds of millions of users to be more productive, so we have a responsibility to get it right.
My time at ICME – especially my teaching - helped me grow a lot, not only as a future engineer but also as a student and a communicator. I think this was critical, because the ability to explain concepts clearly and break them down into understandable pieces is the ultimate key to success, especially in a technical field. I believe being able to communicate effectively will become increasingly important as AI keeps simplifying technical aspects, leaving the ability to collaborate with others the most valuable skill one can have.