What Did You Do This Summer?: Interviews with GRASP Master’s and PhD Students About Summer 2023 Internships

This summer I interned at NASA Ames Research Center through their Pathways program. This internship was fully remote with one site visit.

Carnegie Robotics. It was in Pittsburgh and I had to relocate. Remote work was allowed briefly during the internship.

Zipline – Instant Logistics. I had to relocate to California.

Kodiak Robotics. The internship was in-person in Mountain View, California.

The Fundamental AI Research (FAIR) team at Meta. I relocated to Redmond, WA for an in-person internship.

A GRASP alumnus pointed me to the application because they knew I was interested in NASA. The application was due in April on USAJOBS and required my resume and a short behavior-based questionnaire. I was then contacted by a manager and internship mentor for a virtual interview to see if there was a good fit. There was no technical interview for me.

I first heard about it on LinkedIn and applied through their portal. A few weeks later, I saw that Carnegie Robotics LLC (CRL) was coming to a Penn career fair. I also learned that an alumnus from Kodlab, where I was working, had joined CRL earlier in the year. So, at the career fair, I asked the recruiter if she knew him, and she told me she was the one who hired him. I told her I work in the same lab he used to and we had a brief conversation about it. She then asked that I scan my resume on her iPad and she’ll add me to their list of internship candidates.

A few weeks later, she reached out and scheduled an informative call where she told me what the next steps would be. The following week, I had my first round interview, and two weeks after that I had my second. About a week later, I got an offer from the company.

I heard about the internship through LinkedIn in January of 2023. The application was just resume and basic details.

I saw the position on LinkedIn. The application involved one recruiter screening, one online coding assessment, and two rounds of technical interviews with the engineering team.

I had been on the recruiting email list for Meta for a couple years and was directly contacted by a recruiter after I told my (now) manager that I was interested in interning. I applied around October and was interviewed in January. I had two rounds of interviews, one technical interview and one machine learning/coding interview.

Yes, this internship was allowed thanks to CPT since I’m an international student. The process is not exactly easy. There are a few steps you need to follow and get approval by multiple people. It is complicated, but SEAS (shoutout to Victoria Frew!) has tried its best to make it smooth.

For future students, the administrative department begins accepting CPT applications around mid-April. It is wise to read through all the CPT steps (here) and do whatever you need to do before that date. Once admin begins to accept applications, that approval usually takes 2 weeks, so if you delay it for too long, you could miss your internship start date.

Yes, I participated in UPenn’s CPT as an international student. The CPT process is pretty straightforward and being an international student does not add challenges to the whole process. The only thing we need to take care of is to have our SSN before we start with our internships. So we need to apply for that ahead of time. Other steps can be found on university’s websites which are pretty easy to navigate.

None. I accepted this internship because it directly aligns with my career interests, and it uniquely allowed me to continue working on my dissertation.

I applied to about 60 positions and got calls from 4. I had another offer before CRL to work in Japan. Unfortunately, I had to turn that down because they required that I work for 6 months, and that didn’t fit with my personal or Penn’s timeline. When I got the CRL offer, I was happy with the company, location, work, pay, and timeline.

I applied for multiple internship opportunities. I finalized Zipline because of the work, the amazing team, their product, and the pay, all were perfect.

I applied to north of 100 internships. I also received offers from Shield AI and Apptronik, but Kodiak aligned better with my interests in autonomous vehicles and ML.

I applied to one other internship position. For me, the most important aspects of choosing an internship is research fit and ability to publish; I don’t consider positions that are not directly aligned with my thesis or research interests. There is an opportunity cost that comes with taking an internship (e.g. potentially could delay PhD graduation, or end up working on something that is not interesting to me) so I try to minimize that by prioritizing those two aspects.

For PhD students: If you have a rough idea of your dissertation topic, try finding internship programs/projects that align strongly with your thesis work, and ideally one that will let you continue working on and publishing research related to your dissertation. Generally, a good indicator is if the internship will provide you with additional resources not found at GRASP, but can bolster your thesis (e.g., exceptional computing resources, unique hardware or data, engineering support, etc.). If an internship opportunity comes up that doesn’t fit those criteria, it is still probably an incredibly valuable experience—just be sure to consider the potential delay in graduating if you take it. If you are early in on the PhD and don’t know your topic, there is more freedom to explore different internship opportunities without these considerations. If you are nearing graduation, it might be better to focus on finding internships with the goal of securing employment after your defense.

For MSE students: Given the limited time you have at Penn, I would advise trying to focus on using internships as a networking opportunity. Apply to places you’d consider working for after graduation. Even if you don’t take the internship, getting a face-to-face interview is valuable networking experience. Contacts you make through the application/interview process can be leveraged for future applications for full time positions later.

I recommend building a strong network before your internship search. This doesn’t just mean talking to people at fairs or connecting on LinkedIn. While those may be alright, the strongest networks come from people whom you have worked with – professors, fellow GRASPees at labs, and classmates. Even with those connections, the internship isn’t always guaranteed, but it does increase your chances.

Work on projects that show your technical knowledge and help you learn the pipeline end to end. Showcasing such projects helped me a lot during my internship search. And having clear basic concepts is must. Good coding skills are also required, so keep practicing coding too.

Network with alumni and get referrals. Apply to every place you can find (because you never know). Get your resume reviewed by students who have gone through the same process.

I think the two most important factors of finding and applying to research internships (in contrast to applied/engineering internships) is your previous publications and your network.

Previous publications provide evidence of research fit with your potential team/company and also prove that you can deliver a paper, which is typically the goal of a research internship. Publications also help build your network and reputation within your field.

Because research positions are so limited in industry, it is really difficult to get an interview or offer without already knowing somebody who works there and can give you a strong referral. From what I have seen, the vast majority of PhD interns already had some sort of previous connection or working relationship with someone in the company.

My internship program is unique in that it allows me to continue working directly on my thesis while also contributing to NASA objectives. In that sense, my daily responsibilities did not change much. The only additional responsibilities were training related to using NASA resources and presenting at the weekly or bi-weekly team meetings. My work is a healthy mix of dynamic systems and control, machine learning, fluid dynamics, and aerodynamics.

My primary responsibilities at the company were related to path planning and software development using C++.

My work is related to geometrical computer vision and sensor fusion.

The internship involved creating machine learning models for camera-lidar perception tasks. Day-to-day, it involved keeping up with state-of-the-art research in computer vision, implementing new architectures, curating perception datasets, and deploying new models on the Kodiak self-driving truck.

I work on tactile sensing for dexterous manipulation, which involved a variety of different technical areas: real robot hardware, simulation, machine learning, control, perception, etc.

Yes. I felt well-equipped to make valuable contributions to my assigned project. I also felt like I could speak with authority on topics related to robotics—notable considering most of my office did not have a strict robotics background.

Yes, I had a limited understanding of robotics before I arrived at Penn. The courses I took taught me the fundamental principles of robotics and modern techniques used at leading companies. Through my research assistantship, I learned how C++ is used in robotics. I also learned about designing control loops, typical robotics code infrastructure, and types of sensor data in robotics.

Yes definitely. Specially the courses I took helped me with my theoretical knowledge of the algorithms. My research at GRASP was on a completely different topic of Reinforcement Learning which didn’t translate to my current internship work, but that also helped me learn Deep Learning which would be very useful if I were to work in that domain later.

Yes, a good grasp of machine learning concepts and software engineering skills were essential. Courses and research helped a lot.

I’m glad I gained experience with different types of robot hardware, learned how to conduct research independently, and practiced learning new things quickly.

In an industry environment, it is important to be really good at marketing yourself and your work. It would be nice to have had more preparation for that.

I had an exciting site visit where we conducted experiments in the desert at NASA Armstrong. You can read more about these continuing experiments here.

CRL Golf Friday. The company took us out to a golf course, brought drinks and cigars, and we had a full morning of golf followed by barbecue for lunch.

Learning about the business of drone delivery and all the work that goes on in making the instant logistics work, including both technical and non technical parts.

Seeing my models work after many debugging cycles & hanging out with my coworkers everyday!

There were many other PhD interns at Meta and I really enjoyed making new friends and exploring the hiking trails around Seattle with them!

I gained more soft skills like working on a team with diverse (non-robotics) educational backgrounds. I think this helped me a lot to prepare for a future career in industry where most of your coworkers might not be roboticists.

Yes, at Penn I learned about elementary path planning strategies, like A* search and using occupancy maps. However, at CRL I understood how these algorithms are augmented and polished to achieve reliable results. My internship broadened my understanding of what it takes to make commercial robots viable and safe.

Yes. My internship work was very relevant to the research that goes on in academia too. So I would like to continue working in the same domain and expand my knowledge by taking related courses and doing related research at UPenn. My internship has definitely helped me in gaining skills for my future career in industries. The projects that I have worked on have taught me the real life implementation nuances and problems that would be always useful.

I am a lot more comfortable with the ML engineering workflow, which will benefit me immensely in both industry and academia.

Yes, I got a ton of exposure to many new technical areas and learned how to collaborate with a fairly large, but tight-knit team towards a common goal. This was a really excellent experience and I think it was great preparation for a career in industry.

My internship will continue during summer and winter breaks. We intend on continuing this research collaboration hopefully through the duration of my PhD.

3.5 months. I made connections at CRL that I hope to continue past this summer.

7 months, from May to December 2023. I would be extending my internship to next semester, i.e. in Fall 2023. I for sure made amazing connections in this company with whom I would definitely like to stay in touch for a long time. The people here were very helpful throughout my internship and I would love to stay in contact with them.

The internship lasted 3 months. There is potential for full-time roles, but nothing is set in stone yet.

Six months, from June to November. I am planning to extend my internship to part-time during the fall and will apply for a full time position, if there is one that fits my expertise.

My internship allowed me to continue to pursue my dissertation research full time. Because it was remote, I was fortunate to remain heavily involved in campus and lab activities.

The internship was my only primary responsibility this summer.

I did not do any research along with my internship. I fully focused on my internship.

I did not. 🙂 I focused on full time.

Luckily my internship research will likely directly contribute to my thesis. For ongoing collaborations at Penn, I met with people virtually and worked remotely as necessary, but I focused primarily on my internship.

Yes, absolutely. It has given me valuable insights into what careers at NASA are like.

Yes, it has helped me understand which areas of robotics I’m interested in, what work I really want to do, where I want to be situated after graduation, and what benefits and values I appreciate in a company.

Yes definitely. Since I am aware of the general work that happens in the field I am interning in, I would be able to contribute more in this particular field more in my future job. I also developed skills in this particular area and after exploring this field I also now know that I like working in this particular domain of robotics. So I would be preferring jobs in this domain after my graduation.

I am more sure of my career in software & ML engineering now.

Yes, it confirmed that I really enjoy working on teams with other researchers and that I want to go to industry research after graduation.