Enough Talk- It’s Time to Deliver for Enterprise Data Science
When you’re in an AI-naïve organization, you will invariably hear a lot of noise-a lot of talk on technology, vision, etc. As an Enterprise Data Scientist, you have to go beyond the noise and deliver.
When in an AI-naïve organization, you will hear a lot of noise and talk about how to move forward with data science and AI/ML. And then you find out that these discussions on bringing in AI to the organization have been ongoing for years, and still have no viable way to deploy AI/ML models in production, at Enterprise scale.
Deliverables
You have business buy-in and executive buy-in, and they wonder why it is so difficult to implement end to end AI solutions to solve their business problems. As an Enterprise Data Scientist, they come to you and ask why. So you will tell them it begins with deliverables.
So what do you mean with deliverables? Well, you can’t bring data science capabilities to your organization if you don’t have the foundational documents- the deliverables. And not only do you need these foundational documents, you need to partner with key AI stakeholders in your organization for validation and endorsement of these deliverables.
Environmental Scan
It’s not enough to bring forth these deliverables- they have to be ‘socialized’ into the culture of your organization. Although your executives and the data science team are aligned, you can be rest assured that everyone else will not be. That’s why it’s important to start this initiative with an environmental scan of the organization for data science/AI/ML. In these meetings across your organization, you will start to identify the key AI stakeholders that will help to validate, endorse, and implement the deliverables you were hired to produce. In these environment scans, you will assess their business problems/use cases and discuss possible candidates for AI/data science solutions. For those departments who are new to AI, this is an opportunity to educate about AI and data science. Even if these departments don’t have data science needs now, your education and discussion of what data science is (and is not) will help them to access their needs for the future, and you can be a new contact for them when they have those needs.
In this environmental scan, it will be helpful if you send a ‘data science needs’ questionnaire in advance of your meetings with the various departments in your organization. This way, they will take the time to deep dive into their departments for possible AI use cases, and invite key personnel to answer the questions in the questionnaire. These key personnel would then be present, along with the Director and senior management, in these environmental scan meetings you set up to further discuss. Depending on the size of your organization, this may take up to a year for large enterprises, and a few weeks for small businesses to complete this environmental scan. The most important thing is to not rush- it’s not the answers to the questionnaire that you are after per se. More importantly, you are focused on the socialization of AI and data science into your organization, and your data science team (with executive support/sponsorship) is the one to push this forward. You and your data science team (if you have a team) are also the ones being socialized, as Enterprise Data Scientists are new to them.
Once you complete the environmental scan of your organization, then you can compile and analyze the results. You will review the AI use cases, and then start to cluster the ones that are in the same domain. You will also review their pain points and struggles. Finally, you will get a sense for their needs for data science capabilities.
Data Science Groupings
You will find 3 groups after analyzing the results of your environmental scan:
Departments mature in applied AI research. These groups already have data scientists and are expert at applied AI research and prototyping. They mainly need IT support, governance, MLOps, ModelOps, and infrastructure to deploy their AI/ML models in production.
Departments with interest in data science. These departments have AI use cases they want prototyped, and interested in hiring their first data scientist.
Departments with no current interest in data science. But these groups would like to contact you if they had the need for the future. This can also be FOMO at play, but who doesn’t want to consider AI if it can help with your bottom line?
Regardless of where the groups are, they will all need your services to help them build complete AI solutions, end to end. After the environmental scan, you will have the first two groups as clients, and these are your primary AI stakeholders.
Enterprise Data Science Framework
Now that you have established the need for data science (and data scientists) from the AI use cases gathered from the questionnaire and environmental scan, you can now get to work on deliverables. Your first deliverable will be an Enterprise Data Science Framework for your organization. Previously, data scientists and data science was practiced in siloes. But now you need to help with an Enterprise Data Science approach. This data science framework will be an enterprise document to speak to the need for data science with the AI use cases compiled, and will inform the organization on the specific data scientist requirements for job skills needed to build those complete AI solutions for the AI use cases identified.
Enterprise Data Scientist Roles
With this, the data science framework will help to bring in (hire) Enterprise Data Scientists. The Enterprise Data Science Framework will define the data scientist roles you need. At its basic core, most organizations will need the following:
Business-oriented data scientist (applied AI research and build ML model)
IT-oriented data scientist (build platform and deploy ML model)
These 2 roles are needed to build complete AI solutions, end to end. You need other data professionals, but you need the above two to form the core of the Enterprise Data Science Team, so focus on these 2 roles first. You will need to work with your Human Resources department to help define these 2 roles, and to help them write up job descriptions when posting for new hires. If you already have a department with data scientist needs, then you can partner up with that department as well, and help them to hire the right data scientist for their needs. At the same time, you can test your job description profiles and adjust them accordingly, so eventually, you’ll have validated and tested job postings that can be used by other departments when they need to hire the data scientists according to your framework.
Enterprise Data Science Service Delivery Model
Once you hire the data scientists you need to work on the AI use cases you compiled for the organization, then you will need to define an Enterprise Data Science Service Delivery Model. This Model will help to define how data science services are implemented across the organization. This Enterprise Data Science Service Delivery Model will be delineated in a future article, when I invite my business colleague from my Enterprise Data Science Consultancy to teach you all on our approach and Model.
Other Deliverables
Now that you delivered on your Enterprise Data Science Framework (AI use cases, data scientist roles needed, service delivery model), then you need to continue the momentum from this Framework, and continue to deliver. After the Framework, you need to help deliver on the following, for Enterprise Data Science to take hold in your organization (listed in order of complexity, from least to most difficult):
Plan for external AI Consultants to assist with AI deliverables
AI Stakeholder Mindmap across the organization
MLOps and ModelOps Vision
Data Science Roles and Responsibilities (RACI)
Data Science Best Practices
Data Science Career Path (to attract and retain talent)
Feature Stores and Model Registries
Enterprise AI Project Intake
Enterprise AI Roadmap
Enterprise AI/ML Platform
Enterprise ML Reference Architecture
Data Science Community of Practice and Peer Review
Data and AI Governance
Data and AI Strategy
These are just suggested deliverables. However, if you need help to implement data science at your organization, my Enterprise Data Science Consultancy can assist and customize the deliverables for you (more information below).
At the end of the day, forget about the talk and the noise. If you want to implement Enterprise Data Science at your organization, you need to deliver. Subscribe to this Enterprise Data Science newsletter for more tips on how to implement and practice Enterprise Data Science.
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