Developing AI and ML Practices: Insights from Aaron Jeskey at Bio-IT World 2024

At the recent Bio-IT World 2024 conference in Boston, Aaron Jeskey, Sr. Cloud Architect of PTP provided valuable insights into the development of AI and ML practices within organizations. This discussion focused particularly on the challenges and strategies pertinent to sectors such as healthcare and life sciences.

The Journey of AI and ML Implementation

Aaron began by reflecting on PTP’s six-year journey since its pivotal experience at ReInvent 2018, where the unveiling of AWS’s DeepRacer platform significantly shaped their approach. DeepRacer, a model racecar that teaches reinforcement learning without a traditional remote control, demonstrated the practical impact and educational potential of AI within a business setting.

The Rational AI Architect Framework

Central to PTP’s methodology is the Rational AI Architect, a strategy rooted in applying the FAIR data principles—Findable, Accessible, Interoperable, and Reusable data—in a practical and sustainable manner. Aaron underscored the importance of a grounding statement to anchor any project’s objectives and reduce uncertainty. He cited an age-old adage to emphasize his point: “He who hurries more than he needs to, hurts before he needs to.” This quote encapsulates the philosophy of measured, thoughtful progress in AI and ML endeavors.

Applying FAIR Data Principles

Aaron discussed how each aspect of FAIR data could be effectively integrated from the start of a project:

• Findable: The initial focus is on enhancing data lineage. Understanding not just what the data represents but also its origins and context is crucial for its effective use in scientific research and business applications.
• Accessible: Maintaining data in its most atomic form is essential for accessibility. This approach ensures that even as data undergoes transformations or migrations, its core attributes remain intact and traceable.
• Interoperable: Ensuring data’s usability across different platforms and by various users within the organization maximizes its value and fosters a collaborative environment.
• Reusable: The goal is to keep data perennially relevant. As methodologies evolve or new team members join, reusable data stands as a robust foundation that can continuously support ongoing projects and innovations.

The Importance of Data Lineage and Atomic Data

Aaron elaborated on the practical aspects of data management, highlighting the importance of understanding who provides the data and the quality they can deliver. This understanding helps in making informed decisions about data usage. Keeping data atomic allows organizations to revisit and reutilize their information without the need for costly reprocessing, thus ensuring efficient and scalable operations.

Conclusion: Embracing a Structured Approach to AI and ML

Aaron’s talk at Bio World 2024 illuminated the path forward for organizations looking to integrate AI and ML into their operations effectively. By adopting the Rational AI Architect framework and adhering to the FAIR data principles from the outset, companies can ensure their AI and ML practices are both sustainable and adaptable to future changes and challenges.

For those interested in further exploring these strategies, PTP offers a funded Well-Architect Review that provides deeper insights into applying the Rational AI Architect approach in various environments, ensuring that organizations can embark on their AI and ML initiatives with confidence and clarity.