Do's and Dont's for dataiku implementation

Solved!
amitaryans
Level 2
Do's and Dont's for dataiku implementation

Hi All,

Does anyone have any document which lists the do's and dont's when you are starting with Dataiku implementation.

Thanks,

Amit K

0 Kudos
1 Solution
VinceDS
Dataiker

Hi Amit, 

Here are a few insights when considering a DSS implementation.

A successful DSS deployment should cover 3 aspects: 

1) Projects and Business Outcomes: it is imperative to start with identifying, prioritizing and scoping the projects that will be built in DSS, including their outcomes i.e how they will impact business processes and operational metrics.

> We sometimes use a modified version of the ML canvas to align teams (data, IT, business) around their use cases and make sure we've anticipated all key steps. This, in turn, helps to frame the requirements in terms of training, technical integrations, resource sizing....

2) User Adoption, Training & Governance: based on the projects and desired outcomes, who will be the users that will access the platform and build the projects? What tools and skills do they currently have, and how can they best be onboarded onto the Dataiku platform? Besides core DSS training, how will the users be supported on the long run, both from a functional standpoint (project coaching) and an IT standpoint (platform admin, issues and support...).

Governance and organization around those projects is also key. Defining roles and responsibilities, building an internal center of excellence and implementing sharing & collaboration best practices are key to an efficient deployment.

 > Dataiku provides comprehensive training and coaching programs for users across different profiles, and we're also happy to help by sharing best practices around governance.

3) Technical Implementation: last but not least, the technical implementation stream aims to scope, build and maintain a robust technical environment suited to the achievement of business outcomes. This, of course, includes DSS platform installation and setup, but also all the adjacent components (data sources, execution engines, 3rd party integrations. One key aspect here is also framing the road to operationalization i.e the technical and organizational process by which DSS Design projects will be pushed in production using batch automation or real-time scoring capabilities of the platform.

Hope this helps, happy to dig into more detail if there's a specific aspect you're interested in. 

View solution in original post

2 Replies
VinceDS
Dataiker

Hi Amit, 

Here are a few insights when considering a DSS implementation.

A successful DSS deployment should cover 3 aspects: 

1) Projects and Business Outcomes: it is imperative to start with identifying, prioritizing and scoping the projects that will be built in DSS, including their outcomes i.e how they will impact business processes and operational metrics.

> We sometimes use a modified version of the ML canvas to align teams (data, IT, business) around their use cases and make sure we've anticipated all key steps. This, in turn, helps to frame the requirements in terms of training, technical integrations, resource sizing....

2) User Adoption, Training & Governance: based on the projects and desired outcomes, who will be the users that will access the platform and build the projects? What tools and skills do they currently have, and how can they best be onboarded onto the Dataiku platform? Besides core DSS training, how will the users be supported on the long run, both from a functional standpoint (project coaching) and an IT standpoint (platform admin, issues and support...).

Governance and organization around those projects is also key. Defining roles and responsibilities, building an internal center of excellence and implementing sharing & collaboration best practices are key to an efficient deployment.

 > Dataiku provides comprehensive training and coaching programs for users across different profiles, and we're also happy to help by sharing best practices around governance.

3) Technical Implementation: last but not least, the technical implementation stream aims to scope, build and maintain a robust technical environment suited to the achievement of business outcomes. This, of course, includes DSS platform installation and setup, but also all the adjacent components (data sources, execution engines, 3rd party integrations. One key aspect here is also framing the road to operationalization i.e the technical and organizational process by which DSS Design projects will be pushed in production using batch automation or real-time scoring capabilities of the platform.

Hope this helps, happy to dig into more detail if there's a specific aspect you're interested in. 

amitaryans
Level 2
Author

Thank you VinceDS, this helps.

0 Kudos