Agentic AI in Software Engineering: How and When with Keynote Speaker Vincent Hellendoorn
The field of software engineering is undergoing a radical transformation with the rise of artificial intelligence (AI). AI-powered software agents and large language models are rapidly reshaping the way developers write, test, and maintain code. At the Master of Software Engineering (MSE) program’s 35-year anniversary celebration at Carnegie Mellon University, Vincent Hellendoorn, a leading researcher in AI and software engineering, will provide insights into this technological shift. The event, set for April 4–6, 2025, will bring together alumni, students, and industry leaders for a weekend of discussion, learning, and networking. Learn more about the event here.
Hellendoorn’s keynote, titled “Agentic AI in Software Engineering: How and When,” will explore the growing role of AI in automating software development tasks, from code generation to debugging and beyond.
"Software engineers are already seeing the impact of AI-powered tools, but we are only scratching the surface," Hellendoorn said in a recent discussion. "Large language models are transitioning from simple code completion tools to fully interactive agents capable of handling complex engineering workflows."
His talk will trace the evolution of AI in software engineering, from early predictive models used for code completion to today’s AI agents that can analyze, test, and modify code autonomously. He will also highlight the current challenges and limitations, explaining where human expertise remains irreplaceable.
"AI is a powerful tool, but it's not a substitute for foundational knowledge," Hellendoorn emphasized. "Experienced developers who deeply understand software engineering principles will see the greatest productivity boost. However, for those still learning, AI can sometimes introduce more confusion than clarity."
In this talk below at GitHub, Vincent will review several of these models and discuss their design, impact, and trade-offs compared with GitHub Copilot. Vincent will then take a broader look at the emerging trends in this space and discuss the opportunities and challenges for open source AI tool builders.
Hellendoorn’s research, spanning his time at Google DeepMind and Carnegie Mellon University, has focused on bridging the gap between AI advancements and practical software engineering applications. His work examines how AI can enhance the software development lifecycle, making engineers more efficient while ensuring that automation doesn’t compromise quality.
Looking ahead, Hellendoorn predicts that AI agents will take on increasingly complex responsibilities in software development, including interacting with testing systems, assisting in code reviews, and even participating in architectural design discussions.
"We’re moving toward a future where AI agents will not only generate code but also understand and improve it within a broader development context," Hellendoorn said. "The key challenge is ensuring that these models are both reliable and beneficial for developers at all experience levels."
For students and alumni attending the MSE anniversary event, Hellendoorn’s talk will offer valuable perspectives on how to adapt to the changing landscape of software engineering. His advice to students? Become proficient in leveraging AI as a tool rather than relying on it as a crutch.
"The best engineers will be those who can seamlessly integrate AI into their workflow while maintaining a strong grasp of software engineering fundamentals," he said. "Continuous learning and adaptability will be essential."
The 35-year anniversary celebration of the MSE program provides a unique opportunity to engage with thought leaders like Hellendoorn, reflecting on how software engineering has evolved and where it is headed next. As AI-driven automation continues to reshape the field, his keynote promises to ignite critical discussions on the balance between human expertise and machine intelligence in software development.