Harvard in Tech AI Edition: Eric Lin, Head of Applied AI at Dynamo
Eric Lin, Head of Applied AI at Dynamo AI, has charted an inspiring path to his current role. His journey began during his final years at Harvard, where his research focused on privacy in AI systems. Initially envisioning a career in big tech as a product manager for AI-driven solutions, Eric even worked on projects like Microsoft’s CoPilot. However, his passion for AI safety and transparency led him to reconsider his trajectory. While working on his thesis on AI safety and pursuing a master’s degree at Harvard, he identified the critical challenge in AI development: not just enhancing capabilities but fostering trust and improving safety and privacy.
During this time, Eric connected with Christian Lau and Vaikkunth Mugunthan, MIT PhD students researching similar topics. Recognizing the potential to transition their work from academia to practical applications, Christian and Vaikkunth co-founded Dynamo AI, and Eric became their first hire. Dynamo’s mission focuses on bringing cutting-edge AI safety research into industries like finance and healthcare, where robust safety frameworks are urgently needed. Eric and his team aim to build AI systems that are not only powerful but also secure and trustworthy, ensuring non-technical users have the necessary tools to interact with AI effectively.
Dynamo AI bridges the gap between safety researchers and the everyday risks enterprises face when implementing AI. For example, a bank’s chatbot recommending stocks could create significant legal and financial risks — a practical issue Dynamo is equipped to resolve. By addressing gaps in safety benchmarks and adapting research algorithms for real-world vulnerabilities, Dynamo ensures technology aligns with human needs. Their work involves translating technical advances into actionable safety measures, balancing machine learning optimization with human-centric solutions.
Eric’s deep commitment to the intersection of machine learning research and societal impact has shaped his leadership at Dynamo. At Harvard, he took advantage of interdisciplinary courses at the Business School and Kennedy School, equipping him to move beyond individual technical contributions to leading a team. He emphasizes building trust with users by creating AI that is understandable and responsive. Eric believes that trust is undermined when users cannot discern how AI makes decisions. Dynamo hires individuals passionate about AI interpretability and explainability, recognizing their centrality to fostering a reliable user experience.
AI safety challenges evolve alongside technological advancements. Language and cultural disparities, for instance, present significant risks. While models may perform well in English-centric environments, users in other languages may encounter issues such as spoofing. Similarly, cultural nuances often remain underrepresented in AI design, which can lead to unintended consequences. Dynamo addresses these issues proactively, partnering with organizations like Qualcomm to tackle emerging safety concerns.
In the competitive AI space, Dynamo stands out by prioritizing safety, even when it may seem to slow progress. Eric emphasizes the importance of being transparent with clients about risks. Dynamo focuses on rigorous “red teaming” to stress-test models, offering mathematical safety guarantees and incorporating user feedback to continually refine their systems. This iterative approach ensures AI solutions remain relevant, adaptable, and trustworthy.
Dynamo’s clients include major financial institutions. Eric highlights the importance of building trust through honest communication and practical demonstrations of safety measures. Rather than aiming for perfection, Dynamo prioritizes responsiveness, shortening the feedback loop to ensure models can adapt quickly to changes. Drawing from his research experience at Massachusetts General Hospital, Eric notes that integrating feedback mechanisms is key to winning over skeptics, particularly those in fields like healthcare.
Eric also offers valuable advice to students and emerging professionals. He encourages them to leverage their fresh perspectives to challenge groupthink in AI research and explore how their unique skills can address critical problems. Whether joining a mature company to learn or venturing into startups, Eric advises aspiring innovators to define the impact they want to make and pursue it with purpose.
