What are the best ways to collaborate with machine learning engineers as an Information Architect?

1 Understand the ML problem

The first step to collaborate with MLEs is to understand the problem they are trying to solve with ML, and how it relates to the user needs and the business objectives. You need to ask questions such as: What is the input and output of the ML model? What are the success criteria and metrics? How will the model be trained, tested, and deployed? How will the model affect the user behavior and feedback? By understanding the ML problem, you can align your IA goals with the MLEs' goals, and avoid designing solutions that are incompatible or unrealistic.

2 Communicate the IA vision

The second step to collaborate with MLEs is to communicate your IA vision clearly and effectively. You need to explain how you organize, structure, and label the information, and how it supports the user tasks and goals. You also need to show how your IA design influences the user interface, the navigation, and the interaction. You can use diagrams, sketches, wireframes, prototypes, or other visual tools to illustrate your IA vision, and invite feedback from the MLEs. By communicating your IA vision, you can ensure that the MLEs understand your rationale and expectations, and avoid confusion or conflicts.

3 Define the data requirements

The third step to collaborate with MLEs is to define the data requirements for the ML model. You need to specify what kind of data is needed, how much data is needed, how the data should be collected, cleaned, labeled, and stored, and how the data should be accessed and updated. You also need to consider the ethical, legal, and privacy implications of the data, and how to handle sensitive or personal information. By defining the data requirements, you can help the MLEs plan and execute their data pipeline, and avoid data quality or availability issues.

4 Design the ML feedback

The fourth step to collaborate with MLEs is to design the ML feedback for the users. You need to decide how to present the ML output, how to explain the ML logic, how to solicit user input, and how to handle errors or uncertainties. You also need to consider the user's trust, confidence, and satisfaction with the ML model, and how to enhance or maintain them. You can use principles of user-centered design, transparency, explainability, and accountability to guide your ML feedback design. By designing the ML feedback, you can improve the user experience and the ML performance, and avoid user frustration or dissatisfaction.

5 Test and iterate the ML solution

The fifth step to collaborate with MLEs is to test and iterate the ML solution with real users and real data. You need to evaluate how well the ML model meets the success criteria and metrics, how well the IA design supports the user tasks and goals, and how well the ML feedback influences the user behavior and feedback. You also need to identify and prioritize the problems, gaps, or opportunities for improvement, and propose and implement changes accordingly. You can use methods such as user testing, A/B testing, analytics, or surveys to collect and analyze data. By testing and iterating the ML solution, you can validate your assumptions and hypotheses, and optimize your outcomes.

6 Learn and share the ML knowledge

The sixth step to collaborate with MLEs is to learn and share the ML knowledge with them and other stakeholders. You need to keep yourself updated on the latest trends, developments, and best practices in ML, and how they affect your IA work. You also need to share your insights, challenges, and lessons learned from your IA work, and how they affect the ML work. You can use platforms such as blogs, podcasts, webinars, or conferences to learn and share the ML knowledge. By learning and sharing the ML knowledge, you can foster a culture of continuous learning and improvement, and build trust and rapport with the MLEs and other stakeholders.