AI Tokenomics / AI FinOps SME

Title- AI / Tokenomics SME or Engineer

Location- Roseland, NJ

Type- Long Term Contract

 

Job Description-

Someone who understands AI models and token consumption and can help determine what needs to be tracked, how it should be tracked, and what the different models/consumption patterns mean. This person would provide technical guidance and oversight. . This responsibility is coming under his FinOps/TBM- Technology Business Management organization, and he currently has no dedicated resources to build and manage this capability. AI consumption is increasing across ADP, and Rick needs a more centralized way to understand where AI is being consumed, what it is costing, who is consuming it, and how those costs should be managed and allocated.

 

Key Requirements Discussed

ADP needs to bring together AI consumption information coming from multiple areas, including:

• AI being consumed within ADP products/applications through their AI Gateway

• Developer usage, including GitHub Copilot

• AI/model consumption through AWS/Bedrock and Azure

• Usage information currently being analyzed independently by different business groups

Some consumption data is already available today, but the information and reporting are fragmented across different groups. Rick wants to bring this together into a more consolidated view.

The requirement includes understanding:

• Who is consuming AI/tokens

• Which developers, teams, products and business units are consuming them

• Which AI models are being used

• Token consumption and associated cost

• Which users/models are driving higher consumption

• Opportunities to move to a more cost-effective model, where appropriate

• Centralized reporting / BI dashboards

• Budget and consumption visibility

• Cost attribution and chargeback/showback

• Integration with existing TBM/accounting processes, including capitalization considerations where applicable

• Over time, understanding whether AI consumption is generating the expected business value / ROI

Rick emphasized that the objective is not simply to report an overall AI bill. The consumption and cost need to be connected back to the appropriate developer, product and business unit.