We’re Building the Future of Data Infrastructure

June 26, 2024

Marvell's first GenAI intern: A chat with Ejaz

By Mary Gorges, Talent Brand Content Manager, Marvell

“My advice for high schoolers is to build a strong foundation in math and physics. If computer science courses are available, take them.”

Marvell has always been at the forefront of innovation, and this summer, we welcomed a groundbreaking addition to our team: Ejaz Ahamed Shaik, our first-ever GenAI intern. He’s among more than 400 global interns at Marvell this summer.

Many of us use GenAI but what does someone do whose job title is GenAI? And how did Ejaz prepare for his internship when GenAI arrived on the scene less than two years ago?

Let’s go ask him.

Q: First, how does it feel to be Marvell's first GenAI intern?

Ejaz: It's an incredible opportunity. The job market for software development engineers is now very competitive, especially with advancements in AI. Being at Marvell, where there's a strong focus on innovation, allows me to work on cutting-edge projects and contribute to the future of technology. I do get a lot of attention. There’s a general curiosity among others about what I do.

Q: What are you currently working on at Marvell?

Ejaz: My main focus is developing a retrieval-augmented generation (RAG) system to assist developers with Verilog queries and design issues. (What’s that mean?) This system aims to detect early anomalies in the design process of a chip, which can be costly if not identified before verification. By integrating advanced large language model (LLM) tools, I'm creating a solution that provides accurate and timely insights to developers, streamlining the design and verification process. We're also working on developing a custom large language model specifically for addressing Verilog design and verification issues.

Q: Why is GenAI particularly suited for this work?

Ejaz: GenAI understands and writes code, achieving a level of accuracy that traditional code can't match. It's like having an intelligent assistant that doesn't need to learn languages like Python or Java from scratch. This ability to solve problems using natural language makes it incredibly powerful.

Q: What’s a typical day in your work look like?

Ejaz: A significant portion of my time is dedicated to researching and developing LLM-powered systems. I interact with various teams to discuss the potential integration of GenAI into their workflows. For example, I'm working on developing extensions to Copilot and exploring ways to fine-tune existing models with Verilog data.

 

Q: Why did you choose Marvell for your internship?

Ejaz: My background is in electronics, and I was always passionate about the semiconductor industry. After working as a software engineer and diving into AI and ML, I was looking for roles that aligned with both GenAI and semiconductors and found this role at Marvell, which seemed to be a perfect fit.

Q: Tell us a little more about your background and how, in a short time, you prepared for a GenAI internship.

Ejaz: I have a solid background in computer science, currently pursuing an MS at North Carolina State University, where I’m expected to graduate in December 2024. I conducted an independent study on optimizing code generation using LLMs, working on an AI open-source project called SeedGuard. I also have three years of work experience as a backend software engineer and completed my undergraduate studies at NIT Allahabad, where I worked on a Mask R-CNN object detection model published in the IEEE Journal of Remote Sensing.

Q: What do you think the future holds for GenAI in the general tech industry?
Ejaz: The potential applications of GenAI are vast, and the innovations possible with it are thrilling. For prompt engineers, crafting precise text inputs is crucial to getting accurate AI responses. Techniques like single-shot and multi-shot prompting, as well as using specific phrases like "do it step by step," can significantly enhance AI outputs. Those skilled in these methods will excel in this evolving field.

Q: Can you share some tips for those of us struggling to keep up with GenAI?

Ejaz: I recommend following YouTube tech channels. Some explain tough to understand concepts by being funny and entertaining. Staying updated with the latest developments is crucial, as the field evolves rapidly! Here are some to check out.

https://www.youtube.com/@AndrejKarpathy
https://www.youtube.com/@TwoMinutePapers
https://www.youtube.com/@IBMTechnology
https://www.youtube.com/@statquest

Q. Looking back, how did you know you wanted to be an engineer, and what's your advice for young people?
Ejaz: In high school, I loved mathematics, physics, and chemistry. I wasn't taking computer science courses but was deeply involved in physics and chemistry projects. Watching documentaries about Einstein and Tesla inspired me to dream of becoming a scientist. My brother and I used to conduct crazy science experiments in our backyard, which fueled my curiosity for STEM fields.

My advice for high schoolers is to build a strong foundation in math and physics. If computer science courses are available, take them. Seek out good mentors and talk to people. Internships and projects will help clarify your path. Remember, there's no single right way – I didn't take computer science in high school, and here I am. Enjoy the process and have fun!

Here’s where to go to check out early career opportunities at Marvell: at www.marvell.com/company/careers/university-recruiting.html.



Tags: What Makes Marvell

Recent Posts

Archives

Categories

Accelerated Infrastructure is the Foundation of AI