This summer, Adobe Research interns and mentors teamed up to plan innovative projects, try out new ideas, and share insights that will help shape the interns’ graduate work and push Adobe Research forward. Applications for 2026 summer internships are open now!

To find out more about what really goes on during an Adobe Research internship, we talked with three mentors who were once Adobe interns, along with two of this year’s interns. Here’s what they told us.

This researcher-intern duo teamed up to keep bias out of Adobe’s AI

When Luis Figueroa, Research Engineer, first came to Adobe Research as an intern and GEM Fellow in 2019, he’d just finished his undergraduate degree.

“I think my Adobe Research internship changed my life. I met a lot of brilliant people who lived and breathed research, and that was very inspiring. I was a little intimidated, but my mentors were very supportive. They guided me to find answers rather than trying to just give them to me,” remembers Figueroa.

At the time, Figueroa’s research focused on collecting a dataset and training a model to automatically detect distractions in an image—like people in the background—and remove them. Then, Figueroa joined Adobe Research full time in 2021 and became a mentor himself.

“Becoming a mentor was humbling because I didn’t realize how much time and patience goes into shaping someone’s growth. It’s about trying to ask the right questions to spark curiosity and offering guidance. You’re really partnering. And, even as mentors, we don’t know all the answers and interns come up with things we haven’t thought about. I always find myself learning so much from them.”

This summer, Figueroa worked with Intern and GEM Fellow, Faith Baca, who is beginning her PhD studies in ethical AI development at the University of Southern California this fall.

Baca spent the summer collecting and annotating a large-scale dataset to detect bias and evaluate fairness in AI models. The plan was to cover the most common areas of bias, such as gender presentation, skin tone, and age, and to expand to new ways to assess whether a model is biased. Baca’s goal was to make sure models accurately represent communities that have been historically overlooked.

“As the weeks went on, I realized that there are so many things that I didn’t even think about at the beginning,” explains Baca. For example, she discovered that facial hair was an important element. “After having conversations with different groups at Adobe, it made sense that stereotypes about facial hair can be inherited by models, so it was really important to collect that information.”

By the end of the summer, Baca had created an exhaustive list of labels for evaluating fairness and bias in AI models.

Baca and Figueroa collaborated closely along the way, but there was also plenty of room for Baca to shape her project. “The opportunity to have a leadership role in my own project, while also being guided, was just invaluable,” she says.

Now, Luis is excited about how Baca’s project will impact Adobe products in the future. “Faith’s work will help us identify whether our models are exhibiting bias in any way. If we identify a mistake, we can stop it right there, go back, fix it, and retrain the model so that we can ultimately ship more equitable technologies.”

The eye-opening experience of being an Adobe Research mentor

Research Scientist Vishal Asnani joined Adobe full-time almost a year ago, and before that he spent two summers as an Adobe Research intern. “It was my first-ever experience in the corporate world and I got the chance to work on problems that really mattered,” he remembers. Asnani was pursuing a PhD in content authentication and deepfake detection at the time, and his internship gave him the chance to collaborate with some of the leaders in the field of content authenticity, including his mentor, Adobe Research Scientist Shruti Agarwal.  

During his internships, Asnani focused on content attribution—helping make sure artists get credit for their work. This included collaborating on research papers, developing his dissertation, and helping with the work behind two patent applications. After graduating, Asnani returned to Adobe to continue his research. And this summer, he brought on his own intern.

“Becoming a mentor is an amazing, eye-opening experience,” Asnani explains. “When you’re on the other side of the table as an intern, you’re so focused on solving your own research problem. Now I need to think about someone else’s working style and how they think and then give them enough direction to succeed—while also leaving room for their creativity. I’m trying to replicate what my mentor did for me.”

This summer, Asnani’s intern, Li Zhang, worked on the latest challenges in content attribution. And Asnani notes that, with the increasing complexity of AI prompts and the generated images they produce, the challenges have only gotten bigger since his internship days.

“Interns help us rethink our research and ideas, which is a great exercise. They bring fresh perspectives and they ask so many questions we might not have thought of. Their curiosity pushes us in new directions which we might not have the bandwidth to pursue ourselves,” Asnani says.

Finding the freedom to explore, change course, and think bigger

Research Scientist Jonah Casebeer joined Adobe Research as an intern during Covid. “I was remote the entire time,” he remembers. “And part of what made the internship so remarkable was that I still had such a great time.”

That summer, Casebeer worked on adaptive filtering, the process of removing the distortions and interference that can happen when you’re using transmitters and receivers at the same time—as you do in a video call. It’s a “bread and butter” audio processing problem, Casebeer says, and it felt especially pressing when everyone was meeting over video calls during the pandemic.

During his first summer, Casebeer and his team published research on echo cancellation and by the second summer, they began work on a universal solution for all adaptive filtering problems. It was the first time anyone had approached the problem this way.

As Casebeer explains, his academic mindset as an intern and grad student helped inspire this broader solution: “Lots of times in industrial research, you’re looking to go very deep on one particular problem, but I think a common academic lens is to look at a general case—one that solves not just the problem you’re aware of, but future problems you haven’t encountered yet.”

Now, as a full-time researcher, Casebeer mentors interns of his own, including Dimitrios Bralios, a PhD student at the University of Illinois who just spent his second summer with Adobe Research. Bralios has been working on more efficient ways to train audio-to-audio models.

“Working with really bright and motivated students like Dimitrios is just an absolute joy,” says Casebeer. “I think the biggest change from being an intern to being a mentor is that you have to know how to trust your intern collaborator and know when to focus in together and get into the weeds.”

Bralios began his project with a plan to use a popular new technology, diffusion models. But he quickly discovered that the results weren’t what he’d hoped for—and he was grateful when Casebeer encouraged him to try a new approach.

“We decided to go against the flow of diffusion models and use compressed audio representations instead,” he remembers. “That had a practical benefit for us because it made our experiments smaller with a shorter turnaround time, and we found that you can do a lot more. After making that pivot, a lot of things fell into place.”

This summer, Bralios continued to build on his earlier work, tackling audio watermarking and other ways to identify generated audio content.

“My work at Adobe has evolved into an academic partnership and laid the foundation for my future thesis,” says Bralios.

The experience also gave Bralios a window into industrial research. “I’ve gotten to see how Adobe researchers approach ideas and problems—how they think about things, how they tackle challenges, what they think is important, and which details don’t matter. It has helped me recalibrate my interests and pursue meaningful research. This mix of product and research makes Adobe Research a unique place to be an intern.”

Looking to take your research further? An Adobe Research internship offers the chance to publish at top conferences, work alongside world-class scientists, and see your ideas influence real products. Applications for 2026 internships are open now!