Cerebras Interviewing Guide

Explore

College Careers @ Cerebras

Past Interns @ Cerebras

Learn about our program from past interns...

⁠

Noah A - ML Engineering intern, Summer 2017

How did you come to join Cerebras? I saw Cerebras on handshake and it looked like it had an internship program in ML.

Can you describe the internship? Cerebras had me working on mission-critical problems right away. There were no frills and no handholding. Within my first few weeks I felt like I was making important contributions to the company. At the same time, the people at Cerebras are always very supportive and friendly. I never felt uncomfortable asking for help.

What did you learn? In addition to learning about ML hardware, I learned a lot about how startups work. Andrew, the CEO, spent a lot of time with the interns and made sure we understood how the company operates.

What is your lasting impression of Cerebras? Cerebras is a fast-moving company full of smart, serious people tackling difficult problems. As a full time employee, I continue to be challenged and learn every day. I don't think I could have chosen a better company to start my career with.

⁠

Jay J - Integration Engineering intern, Summer 2020

How did you come to join Cerebras?

I had no idea about Cerebras, it was at a dining table discussion where one of my

friends casually mentioned about Cerebras Systems as we were discussing some great

work going on in the AI/Chip industry. I got interested in and visited Cerebras website

and happened to apply to the open intern roles.

Can you describe the internship?

In my opinion, the company is now undergoing standardization and systematic

restructuring as we are scaling up and catering to a spectrum of current and potential

customers. The USP of the company is the people of Cerebras, led by Andrew Feldman

and Sean Lie, it’s a fulfilling experience to work alongside other talented engineers and

research scientists.

What did you learn?

Being a part of the Integration Team helped me understand TensorFlow estimators and

how different models are taken from concept to code and finally running it on the

Cerebras Systems’ CS1. Apart from the projects I worked on, I got insights about

our customer models such as ANL and their research interests.

What is your lasting impression of Cerebras?

Great co-workers, good work environment and currently at an inflection point of potential

growth

⁠

Prithvi G - Kernel Engineering intern, Summer 2019

How did you come to join Cerebras?

I found out about Cerebras through a job posting on Handshake in my junior fall. There actually was not a whole lot about what exactly Cerebras was doing at the time. The general idea I had was that they were working on some kind of hardware accelerator for deep learning. The interview was a great exercise that helped me understand the type of problems that the Kernel team solved and gave me a better sense of what I would be working. When I then talked to Sean (founder), he got me really excited about what the company was working on, especially since I was very interested in heterogenous computing, domain specific hardware, and machine learning at the time. I was also really impressed by how quick the turnarounds were in the process. This was in stark contrast to some of the bigger companies like Google that I was dealing with at the time.

Can you describe the internship?

My internship lasted for 13 weeks, and I worked on two projects that took roughly six weeks each. The projects were very much relevant to what was being worked on at the time. However, at the same time, they were relatively self-contained, so I got to see them through from start to finish. The internship ended with a presentation to Andrew (CEO), Pierre (investor), and the rest of the interns.

What did you learn?

During the internship, I got a better understanding of some of the technical challenges related to the problems that I was working on. At the same time, it was interesting to be exposed to the types of problems that other people were solving in completely different areas through the 4-slide talks.

What is your lasting impression of Cerebras?

I think what made me want to return to Cerebras was that the things being worked on aligned perfectly with my interests. I also really enjoyed working with the people that I worked with last summer, especially on the kernel team.

⁠

Vijay T - Kernel Engineering intern, Summer 2020

How did you come to join Cerebras?

I first heard of Cerebras during August 2019 when I was following HotChips. This was the first time Cerebras had disclosed technical details of the WSE. Ever since then, I was captivated by the idea of working on the world's largest processor. So when I visited the Cerebras booth at SC19 and saw the CS-1 in person, I was compelled to apply online.

Can you describe the internship?

Each layer/kernel has potentially millions of possible configurations in which it can be mapped onto the fabric of PEs, and finding the best one during model compilation is critical for performance. This must also be done as quickly as possible. One of the critical parts of this process is predicting how much maximum memory the kernel will consume. The most attractive part of this problem for me is just how unique of a software-hardware co-design project this is. On one hand this is similar to regular code compilation, but on the other, the process more closely resembles a PandR process from traditional VLSI/FPGA workflows.

What did you learn?

This was my first time using many of the tools I used during this internship, namely data science tools such as Pandas and ML frameworks like Pytorch. I also got to think about incredibly difficult co-design problems that helped shape my future research direction. This being a fully remote internship, one of the biggest self-developments was being able to work in such a asynchronous and distributed fashion.

What is your lasting impression of Cerebras?

A company looking to solve more world's first problems than I have ever seen before. It is an incredibly humbling place, because of both the technical challenges involved and the brilliant people working to solve them.

Want to print your doc?
This is not the way.

Try clicking the ⋯ next to your doc name or using a keyboard shortcut (

CtrlP

) instead.