AI Interns: Making a Difference at Intel

Artificial intelligence is evolving in transformative ways all around us. Today’s young leaders in AI will help to determine its future. Inventing AI applications that benefit everyone requires that more people have opportunities to participate in AI’s evolution and governance. With this in mind, Intel offers several career programs, including internships, to get AI technologies into the hands of more people and to enable the next generation of data scientists and the world-changing innovations they will create.

AI Internships at Intel provide a unique opportunity for students to advance their academic research projects and gain professional experience at a company that provides an unmatched solution portfolio for a wide range of AI applications. Students also often find that their Intel internship expands their perspective on the wide variety of opportunities for careers in AI, prompting them to investigate job openings and career directions that would otherwise go unexplored.

Ultimately, AI internships at Intel foster connections. Interns might meet a new advisor that helps secure funding for their project, co-author a research paper that is presented at a major AI conference, or even continue their academic research through a career at Intel or another institution.

Here’s what our student interns have reported to us about their experiences at Intel.

Shauharda Khadka, Data Scientist at Intel AI

What are you studying, where are you studying it, and what is your progress there?

I am currently a data scientist at Intel. However, prior to converting into full-time a couple of months ago, I was a Robotics Major at Oregon State University. I joined Intel last June and interned for about a year before making the transition.

Why did you seek out an internship with Intel? What made you excited to intern here?

I chatted with a couple of researchers at Intel and was excited by the breadth of research Intel was conducting in Deep Reinforcement Learning. This aligned with my interests and I sought to join Intel to be a part of this team.

Figure 1: High-level schematic of Collaborative Evolutionary Reinforcement Learning

Figure 1: High-level schematic of Collaborative Evolutionary Reinforcement Learning

What projects are you working on?

Research and development of an AI-driven controller for controlling a 3D musculoskeletal model of a human body with a prosthetic foot. The goal was to inform designs for real-world prosthetics. We presented a workshop paper on AI and Prosthetics at NeurIPS 2018.

Research into Collaborative Evolutionary Reinforcement Learning (CERL) – a hybrid method that combines policy gradients with ideas from natural evolution towards improving sample efficiency in reinforcement learning algorithms. We presented a paper on CERL at ICML 2019. Now, we’re using CERL to improve kernel mapping in Intel® Nervana™ Neural Network Processors to make them faster for inference workloads.

Research into Multiagent Evolutionary Reinforcement Learning (MERL) that extends the hybridization of evolution and policy gradients to tackle difficult multiagent reinforcement learning settings. This paper was accepted at the Deep Reinforcement Learning Workshop at NeurIPS 2019 and is under review for ICLR 2020.

Huili Chen, PhD Student at the University of California, San Diego

What are you studying, where are you studying it, and what is your progress there?

I’m a fourth-year Ph.D. student majoring in Intelligent Systems, Robotics, and Control at the University of California, San Diego.

Tell us more about your internship. What made you excited to intern here?

I’m working with Intel on secure and reliable machine learning during my Ph.D. studies. I’m excited to discuss the state-of-the-art progress in the machine learning field with the talented people at Intel and learn from them.

I’m working on developing an end-to-end compiler framework for privacy-preserving machine learning (ML) inference, in particular, a framework that automates and accelerates homomorphic encryption (HE)-based ML computation. I finished the prototype of the HE compiler framework and it is able to accelerate common ML operations, including matrix multiplication and convolution. The framework I developed achieves faster runtime compared to the standard implementation. I am writing a conference paper on this topic together with my manager, Rosario Cammarota.

Intel provides great learning materials and computing resources, which is very helpful. I plan to extend the current framework to enable complete inference of a neural network for my project.

Following your internship, what’s next for your education and career?

The internship at Intel has helped me get a deeper understanding about privacy-preserving machine learning. This has broadened my research interests and motivated me to build secure and private ML systems during my Ph.D. program.

Landan Seguin, Deep Learning Data Scientist at Intel

What are you studying, where are you studying it, and what is your progress there?

I am studying computer engineering at the Georgia Institute of Technology (Georgia Tech), and I was in my senior year (I graduated at the end of the summer).

In a few sentences, please tell us more about your internship at Intel.

I was tasked with defining my own research project in the field of deep learning. I decided to develop methods that would improve the labels used to train image classification models. The potential benefits include increasing the model’s accuracy, adversarial robustness, calibration, and data efficiency. The results are still inconclusive, but I plan to continue the research during a portion of my time.

At Intel, I am able to interact with many great researchers who provide invaluable feedback. All of the researchers are very kind and open to discussion; it is an amazing community.

Following your internship, what’s next for your education and career?

I accepted a full-time offer at Intel which was provided to me during my internship. I will be on the same team, but I will be working on a more customer-focused project. My work from my summer internship may be useful to my new project and I am hoping to dedicate a portion of my time to completing the research.

Karan Mahesh, Sophomore at the University of California, Berkeley

What are you studying, where are you studying it, and what is your progress there?

Currently, I’m a second-year mechanical engineering and math major at the University of California, Berkeley.

Why did you seek out an internship with Intel?

I really wanted to work for a technology company that had its hands in many different industries so I could get a better understanding of the general trends of big tech companies. Also, I wanted to expand my knowledge and understanding of artificial intelligence and its applications and Intel’s AI push was new and exciting to me.

What projects are you working on?

This summer, my main projects were to improve the Neural Network Distiller, a deep learning optimization suite that Intel had developed, and to develop some of the framework for a hardware-aware quantization system.

Moin Nadeem, Senior at the Massachusetts Institute of Technology (MIT)

Why did you seek out an internship with Intel?

My roommate was interning at Intel and loved it. I loved the autonomy they gave me and the ability to make a difference upon the organization.

In a few sentences, tell us more about your internship at Intel.

I worked as part of the AI research team on a project on de-biasing natural language processing (NLP) models. We’re publishing a paper on November 1 with some really interesting results!

This wouldn’t be possible without Intel and Anna Bethke. Intel did everything from fund the creation of the dataset, to provide compute, to enabling me to release the work!

Following your internship, what’s next for your education and career?

I am starting my Master’s in this area, and because of Intel, I found an advisor who was interested in my work and decided to fund me (i.e. no loans for my Master’s). This is incredible, and my internship provided me the experience to do that.

Siam Umar Hussain, PhD Student at the University of California, San Diego

Why did you seek out an internship with Intel?

I was familiar with the research done by the Intel AI team for the past few years. They are working on some exciting projects that closely match my research at UCSD. Moreover, I have known my manager Ro and his research for about 4 years. Working with him was one of the main motivations of joining Intel.

Tell us more about your work.

My project is on the scalable execution of Secure Function Evaluation (SFE) through the Yao’s Garbled Circuit (GC) protocol. Right now, we have finished building the GC framework. A publication on the enhancements by this framework over the state of the art is currently under review. We are working on developing some exciting privacy-preserving applications with this framework.

In my opinion, the biggest perk of working at Intel is learning from some of the most talented researchers in this field. Moreover, working at a large organization like Intel gave me a broader overview of the current state of the art and potential direction of future research in my field.

Following your internship, what’s next for your education and career?

I am currently in my final year of my PhD program and my project at Intel is going to be a significant part of my thesis. I have made tremendous progress towards my degree during my internship.

Take Your Next Steps in AI

Intel hires hundreds of interns each year. Some go on to careers at Intel AI. Others find AI-focused roles within teams in other areas of Intel. Others still return to academia or find other roles within the AI community. Regardless, we are proud of Intel’s support for these future AI innovators and the contributions they make to this field.

If you’re an undergraduate, graduate, or PhD student interested in a career in AI, we encourage you to get in touch with us via our AI Internships interest form, explore the possibilities of AI careers at Intel, and hone your AI skills with our Intel® AI Academy.