Maritza Ramirez Mills lives in the beautiful state of North Carolina and graduated from the OMSCS program in May 2022. Maritza currently works as a product manager for Google leading part of their AI/ML growth strategy. Her focus is on continuously improving the experience of customers running HPC (high performance computing) workloads on Google infrastructure. She also partners with engineering, research, and go-to-market teams on identifying new applications for machine learning that Google can make available to their customers.
Maritza enrolled in OMSCS because she was interested in pursuing more advanced topics in computer science, especially machine learning and AI. Prior to OMSCS, she was enrolled in an on-campus program but found that an in-person program wasn’t compatible with her professional life. She spent a lot of time commuting and trying to fit class schedules around her work schedule. What finally inspired Maritza to apply to OMSCS was realizing she needed to relocate for a new job. OMSCS gave her the flexibility to do her coursework on her own schedule and allowed her to move across the country two different times while still being able to complete course work. “Whenever I switched jobs, I had the flexibility to pause my coursework for a semester and focus on getting up to speed, then later blend my coursework into the rest of my work and personal life.”
So, what does Maritza love most about OMSCS? In her own words:
There’s a lot to love about OMSCS, the rigor, the flexibility, the price; but what surprised me the most, and what has left the most lasting impression, is how connected I ended up feeling to the rest of the OMSCS community and all the great friendships I was able to build in the program. Being an online program, I assumed I’d go to class, get my work done, and there wouldn’t be much time to interact with other students. I was wrong. I met a lot of other students through class, group projects, and the online Slack community who I’ve continued to keep in touch with. We commiserate about school projects, support each other’s careers, and sometimes meet in person when the circumstances allow it. Just last year OMSCS launched its inaugural in-person conference. It was a great way to meet professors and other students in the program: students who already graduated, students who had just been accepted, and those still slogging through it. There were students there who flew in from all over to attend. Everyone I met was curious, supportive and open to learning from each other. The community environment that OMSCS has fostered is truly unique.
I’m sure I’m not alone in saying that Dr. Joyner has been one of my favorite professors in the program; I’ve taken several of his classes. I especially enjoyed the way he structured CS 7637: Knowledge-Based Artificial Intelligence. For me, it’s the perfect blend of programming, writing, analysis, and learning from peers. You get hands-on experience building your own AI agent, while also having the opportunity to take a step back and explore the big questions in cognition. There is also a semester-long course project that allows you to review other students’ approaches between milestones and tailor your approach based on what you learn from your peers. This experience really fosters a sense of community between classmates. My favorite class was CS 7641: Machine Learning with Dr. Isbell. The projects in Machine Learning were open ended, forcing you to think carefully about how to structure your experiments. I got a deeper appreciation for why building an effective ML model is so difficult, and many of the pitfalls you can run into when interpreting results. And for future students considering the course, I would be remiss if I didn’t mention just how entertaining the class lectures are. I personally enjoyed the witty banter and found that the way Dr. Littman and Dr. Isbell discuss the course material often made lectures feel more like listening to a podcast.
In terms of research:
When I was in the program, I spent time on a self-directed project related to AI creativity and its relation to human creativity. My team and I evaluated different methods for building computational music models, and then we compared our models with the different paradigms for human creativity from history through modern times. I’d love the opportunity to continue studying this topic. One question I have is about how generative music models will impact songwriting. We know that recommendation systems already influence how people find and discover music. I’m curious how those same recommender systems might influence the creative process. For example, some artists write music from scratch, but they might also choose to use samples. Those samples may include a mix of both AI and human-generated music. I’m also curious about how AI could help influence the lyric writing process. AI could help songwriters find words for the music they’ve written. Imagine an application that could help people find words with different syllables but similar meaning or emotional range so that it still fits the context of the song. Conversely, people could write song lyrics and then use those lyrics to search for music samples that help them express those words through music. Those samples could be labeled by their most basic building blocks: rhythm, beat, tone, melody, etc. and then categorized by mapping the associations of each sample’s attribute to lyrics in other songs, or by their emotional range, for example. I’ve noticed a lot of activity in this area within the startup ecosystem and would love to pursue these questions in one way or another.
In fact, if Maritza could collaborate with any person, past or present, it would be:
Margaret Boden and Daniel Levitin. I have learned so much from their work and would love the opportunity to collaborate with them. Boden influenced much of the early philosophy on artificial intelligence, especially AI creativity. I referenced her work extensively during my investigation of generative music models. One insight that stood out to me was from her paper “Creativity and Artificial Intelligence” which outlines three types of creativity: combinational, exploratory, and transformational. Combinational meaning mixing familiar ideas in a new or unfamiliar way. So, I like to imagine someone doing a cover of your favorite song, but within the context of a new genre. Using these definitions, my group and I were able to argue that generative music models do demonstrate a sort of combinational creativity. During our project, I also had the opportunity to read Daniel Levitin’s book This is Your Brain on Music, Which provided a foundation for understanding the impact of music on people and how their bodies respond to it. A common component in many of the perspectives I came across for creativity was this idea that the creative output must be considered valuable, and usually that value was determined by its impact on a human evaluator. Levitin’s research pointed us toward methods for measuring the physiological response that music has on people and gave us examples of how we could structure future research on this topic.
Furthermore, Maritza has plans to continue studying areas related to cognitive science:
Cognitive science is an interdisciplinary field that lets you study the mind from the perspectives of computer science, philosophy, psychology, anthropology, and more. Technology has a way of changing the way we see ourselves and relate to each other. For example, the capacity for AI to create artistic outputs has caused a lot of us to question what exactly makes a human being creative, and what is unique about human creativity. I spent some time digging into this question during the program and learned that our concept of humans as intentional creators is a relatively modern perspective. For example, there are older notions that creativity was something that occurred in nature and that historically, humans saw ourselves as being inspired or skilled imitators of those external sources. Even in the modern perspective, creativity is often the result of learning from the creative work of others, and people do not create artwork that is isolated from the influence of their environments. Understanding that history provides more context on how computationally-generated art might affect our identities as creative beings. As part of my work as a product manager, I’m required to deeply understand how what we’re building is going to impact our customers. Continuing an interdisciplinary education would allow me to further explore what impact a given technology is going to have on people and understand how our culture influences the direction of technological advancement. In the previous example, AI-generated art has also raised concerns about the impact this will have on artists and creative ownership. One of the most immediate responses from the US legal system was that copyright protection only applies to human-made artwork. This is just one example of how our societal values impact the way technology is distributed and monetized. With the diversity of classes offered by OMSCS, I’ve been able to pursue some of these interests in classes like Knowledge-Based Artificial Intelligence, CS 6603: AI, Ethics, and Society, CS 6750: Human-Computer Interaction, and CS 6795: Intro to Cognitive Science. I’m really thankful for that exposure.
How does Maritza relax after such incredible work? She is a vocalist who loves to spend time discovering new music and attending live music shows, especially of up-and-coming artists. “One of my favorite past times is reviewing song lyrics and discussing them at length with friends and family.”