To Scale and Deploy your AI/ML Models, first Scale and Align your AI Stakeholders
The bottleneck to Enterprise Data Science is not the technology- it's the people. I used to think this was cliché, but I found out the hard way. Align the AI stakeholders before building the platform.
I have a confession- I tried to do this backwards. In my zeal to bring Enterprise Data Science to my organization, where data science was previously siloed and had no Enterprise Data Science stack, I focused on architecting and building the Enterprise Data Science Platform. I thought for sure that if we build it, they will come.
However, despite other IT leaders telling me to slow down and focus on people, I focused on the technology stack and my fancy new frameworks of MLOps and ModelOps. But the more I pushed on the AI prototypes to be run through our Minimal Viable Platform, and instill MLOps and ModelOps concepts to both focus on the seamless development and deployment of AI/ML models, while instilling Continuous Integration and Continuous Delivery at Enterprise scale, the more push-back I received from various stakeholders. I was confused- isn’t this what they wanted?
Fortunately, I had good mentorship, from Data Scientists /Enterprise Architects who have been in my same position before. They told me ‘people before technology.’ And when I didn’t heed their advice, they sat patiently until I returned a few months later to say they were right. It’s all good though, as it taught me a valuable lesson: to heed advice, no matter how cliché or trivial they sound at first glance.
Without engagement and alignment from key AI stakeholders across the Enterprise, then the Technology platform you built for the Enterprise has a good chance to become an orphan platform. When AI is previously siloed, and when there are no centralized guidelines for AI, then there is need to govern AI at Enterprise. Without AI governance, you can’t scale and deploy your AI/ML models at the Enterprise. You first need to scale and align your AI stakeholders.
When I first attempted to find the various stakeholders, most were doing their own:
AI governance,
AI vision
AI roadmaps,
AI development guidelines,
AI project intakes
local AI/ML ‘technology stacks’
AI prototypes
AI/ML local ‘deployments.’
So Data Science was being performed mostly solo or in small teams, and each of the dozens of AI teams sprinkled across the organization ascribed to many of the above for their own siloes, contributing to duplication and burnout or departures of data scientists due to their models not being able to deploy at Enterprise scale.
As the AI/ML Tech Lead for IT, I recognized the need for IT align with the key AI stakeholders, including:
AI governance and AI project intake (business or data organization)
AI ethics (business or data organization)
Business SMEs (business lines)
Applied AI Research, AI development guidelines, and MLOps (business or data organization- functional sponsor)
Enterprise AI/ML platform (business or data organization- functional owner)
Enterprise AI/ML platform and ModelOps (IT- technical owner)
Because I am leading the technical platform from IT, my IT team and I have to align with all of the other AI stakeholders from items #1 to #5. For a technologist, focusing on people over the code and stack is a major pivot in the development of Senior Individual Contributors to AI strategist/leader. This development and pivot will be fast-tracked, as I help my IT team to align with the other 5 AI stakeholders.
For the takeaway message, in order to scale and deploy your AI/ML models at Enterprise, you need to scale and align your AI stakeholders. And never forget:
People over technology
If you're looking for support from me, here are a few options:
Enterprise Data Science Consultancy: With my consult team comprised of a Senior Data Scientist, Senior ML Engineer, Senior Data Engineer, and Senior Cloud Engineer, we will help you architect and build your Enterprise Data Science platform, and transfer knowledge to your IT team to maintain and optimize it. We will also overlay an MLOps framework to manage the AI solutions you build on this platform. If you don’t have an MLOps team, we will help you build one. Please get in touch about this consultancy here
Coaching and Mentorship: I offer coaching and mentorship; book a coaching session here