Making models that adapt: PhD student Christopher Clarke on personalized AI and his Guyanese roots

Recent University of Michigan PhD grad Christopher Clarke is on a mission to make AI models that are personalized and impactful, while also fostering tech growth in his native Guyana.

What does it take to go from a fledgling programmer in Guyana to an AI trailblazer? From a high school programming course in his home country of Guyana to completing his PhD in computer science and engineering at the University of Michigan, Christopher Clarke has traversed an inspiring educational path, driven by a passion for technology and a commitment to his roots.

The road to research

Christopher Clarke headshot
Christopher Clarke

As Christopher reflects on his academic journey, he credits his initial interest in technology and computing to his high school days in Guyana. He got his first taste of programming through a school-based project using Pascal, an old programming language that introduced him to the world of computer science.

“I’ve always been interested in technology,” said Christopher. “In high school, I got my first formal introduction to programming, and I found it interesting enough to learn more on my own. I even later returned to my high school as an IT teacher.”

From there, Christopher completed his undergraduate studies in computer science at the University of Guyana and was so successful that he subsequently joined the faculty as an instructor. In addition to his teaching, Christopher was also involved in entrepreneurial endeavors, launching a startup called V75 Inc. that focuses on developing innovative software solutions and providing tech services to address local and regional challenges in Guyana.

Christopher’s journey to the University of Michigan began with a meeting with CSE Professor Jason Mars, of Guyanese heritage himself, who visited the University of Guyana as part of an educational initiative. This serendipitous meeting completely transformed Christopher’s trajectory.

“A PhD wasn’t in my sights initially,” said Christopher. “But after meeting Jason Mars and collaborating on a few projects, it wasn’t long before I found myself in Michigan.”

Transforming AI through personalization

Adjusting to the PhD program at U-M was not without its challenges. Coming into a completely new environment was a big leap, both academically and culturally. And while the cold weather can still be a struggle, Christopher has found U-M to be home to immense resources and opportunities that have allowed him to flourish and find his niche in the research world.

A photo of Christopher Clarke typing on a computer from behind. On the screen are lines of code against a black background.
Christopher Clarke works in the lab in the Bob and Betty Beyster Building on U-M’s North Campus.

“I came in at a time when the lab was kind of in a reset mode,” said Christopher. “This gave me the freedom to pursue my own research interests. I started independently, taking classes, exploring various facets of computer science, and honing in on my unique areas of interest.”

This sense of freedom throughout his PhD program allowed Christopher to explore numerous areas and carve out a space for himself in the field based on his own interests, rather than adopting his advisor or labmates’ specialties as his own.

He first dipped his toes into the research waters with a collaborative project with Ford involving conversational AI agents in vehicles, an endeavor that would lay a foundation for the rest of his PhD journey. 

“We worked on developing a system that could interact with drivers conversationally, enabling them to control various vehicle functions through voice commands,” Christopher explained. “It was my first introduction to AI and it set the stage for my future research.”

From this initial entry, Christopher’s research interests have expanded to span several areas within AI, particularly in natural language processing (NLP) and human-AI interaction. 

A core focus of his current work is personalization in AI models. Christopher argues that while increasing data improves AI performance, it is crucial to create models that can tailor their outputs to the unique needs and experiences of their users.

“Models like ChatGPT are incredibly powerful, but they often lack the ability to adapt to individual user needs,” he said. “We need models that adapt to individual users’ knowledge and perspectives, rather than generalizing from large data sets.”

To this end, Christopher has spearheaded the development of various techniques for personalized model training, allowing AI systems to better respond to specific user needs. For example, his current work on parameter-efficient tuning for personalization explores how models can be fine-tuned using minimal data to reflect individual user preferences effectively.

Another notable contribution is Christopher’s development of Rule by Example, a framework aimed at improving AI’s ability to detect hate speech. By creating a feedback mechanism where users can correct and guide the model’s learning process, the system becomes more attuned to individual and community standards of harmful content. This approach not only enhances the accuracy of hate speech detection but also makes the model more controllable and explainable.

It’s essential to find something that grounds you and aligns with your values. It’s about finding the balance between theoretical research and practical, impactful solutions.”

Christopher Clarke

Additionally, Christopher has developed datasets and benchmarks essential for personalized AI. He introduced the GuyLingo corpus, the first dataset for Guyanese Creole, emphasizing the importance of including diverse languages in AI development. This project highlights Christopher’s commitment to making AI inclusive and representative of various cultures and languages.

Christopher has published his research on these topics at renowned conferences, including ACL and EMNLP. He has also collaborated with industry leaders including Microsoft, Amazon, Ford, and Clinc, contributing significantly to both academic and practical advancements in AI. 

Making an impact at home

In addition to his academic and research pursuits, Christopher is passionate about fostering technological growth in his home country of Guyana. He has launched two startups dedicated to advancing the Guyanese tech sector, V75 Inc. and Nexus Hub Inc.

V75 Inc. is a for-profit company that delivers custom software development, IT consulting, and digital transformation services, with the aim of driving technological adoption and efficiency across various industries in Guyana, including education, healthcare, and finance.

Nexus Hub Inc. is a non-profit organization that aims to bridge the digital divide by upskilling young people in technology. Through this initiative, Christopher and his team facilitate courses, apprenticeship programs, and community development projects, all with the goal of empowering the next generation of tech leaders in Guyana by providing them with the knowledge, skills, and mentorship needed to succeed in the tech industry.

Christopher Clarke holds a microphone and speaks while standing on a stage at the front of a large room. Behind him is a projector screen with the title of the conference on it.
Christopher delivered a keynote talk at the 2023 AI-4D Showcase in Georgetown, Guyana.

Through his entrepreneurial work, Christopher seeks to facilitate collaboration and support innovation in the tech sector, helping Guyana achieve greater economic stability and growth.

“We need to grow our own tech community in Guyana,” he explained. “Particularly with the country’s newfound prominence in oil, it’s crucial to diversify our economy and have skilled individuals who can innovate and solve local problems.”

Future pathways

As Christopher prepares to transition from his PhD, he sees a future where he balances both research and practical implementations of technology. He has gained exposure to both academia and industry through internships at tech companies like Microsoft and Amazon, and through teaching and mentoring at U-M.

“I haven’t fully committed to a single path yet, but I know it will be a blend of research and developing real-world systems,” he said. “I believe in building impactful technology while also tackling interesting research challenges.”

Christopher emphasizes the importance of having interests outside of academic research to maintain a balanced perspective. He advises future PhD students to deeply consider their motivations for pursuing a PhD and to seek a supportive environment and advisor.

“Michigan has provided me with all the resources to succeed, but it’s essential to find something that grounds you and aligns with your values,” Christopher noted. “It’s about finding the balance between theoretical research and practical, impactful solutions.”

From his beginnings in Guyana to his groundbreaking research at the University of Michigan, Christopher’s journey is a powerful demonstration of how dedication and innovation can drive meaningful change. Whether advancing AI research or nurturing the tech community back home, his work has already left a lasting mark in the field of technology and beyond. As he embarks on the next phase of his career, Christopher is poised to continue making significant strides in AI, with a steadfast commitment to creating meaningful, real-world solutions.