Job Title: Data Analyst / Data Architect

Contract : 6+ Months (Remote)

Job Description:

Purpose  

Turn inconsistent raw sources into one coherent model that answers who spent what, on which application, for which purpose. This role does the analysis that must happen before engineering can build: profiling what each source actually provides, mapping it to the canonical schema, and designing the transformations that produce the emergent data on which transparency, accountability and optimization depend.  

Key responsibilities  

• Profile raw data landed from assigned tool integrations — gateways, observability platforms, productivity tools, AI-enabled SaaS — and establish which usage, identity, license and cost fields are genuinely available rather than assumed.  
• Analyze hyperscaler cost and usage data, including AWS CUR 2.0 with caller-identity allocation, Azure and GCP billing exports, and the extraction of model metadata from SKU and description attributes.  
• Design silver-through-gold transformations for assigned sources, documenting the design and the gaps needing a fallback, so engineering builds from a specification rather than discovering the shape midbuild.  
• Contribute to the canonical schema and its alignment to the FOCUS billing specification.  
• Help design and document the attribution precedence — resolving each record through caller identity, gateway telemetry, observability data, resource tags or account tags in a defined order, with the mechanism differing between direct attributes and usage-based allocation.  
• Design allocation logic that splits cost billed to a shared endpoint across its real consumers, aligning telemetry token counts to billed token counts at a common grain of model, token type, tier and period.  
• Analyze the current application identifier population, ranked by spend, assessing each on two independent axes: whether the tag is correct, and whether the endpoint is single-purpose or shared.  
• Analyze ServiceNow business application records to determine which attributes can classify an application as internal or external revenue-generating, including how completely those attributes are populated.  
• Support the taxonomy derivation rules that place every dollar on two axes — audience and environment — and the coverage measures reporting how much cost can actually be placed.  
• Design reconciliation between observed usage and vendor invoices, with a published variance tolerance and a defined home for the residual.  

Essential skills and experience  

• Interest and aptitude to dive in a really learn the data.  This is not a black box exercise,  the successful candidate will be determining the key cross references to join disparate data sets, merging tool-based usage data with billing data to provide a combined cost and usage story.   The scope includes many net new data sets that do not have clear documentation or subject matter experts to guide the work.   
• Strong data profiling and exploratory analysis on messy, semi-structured vendor data — API responses, JSON payloads, billing exports — using SQL and Python.  
• Dimensional modeling to a professional standard: conformed dimensions, slowly changing dimensions, and effective-dated mappings where a person or application changes ownership mid-period.  
• Excellent written documentation. Much of this role's output is specification others implement; ambiguity in the document becomes divergence in the build.  
• Able to elicit requirements from tool owners with other priorities, and to reach a workable answer where the ideal data is not available.  

Desirable  

• FinOps practice experience, or cost allocation in a regulated or multi-divisional enterprise.  
• Familiarity with CMDB and service hierarchy data structures.  
• Exposure to LLM usage telemetry — tokens, prompts, model versioning, gateway logs.  
• Experience designing allocation, showback or chargeback models where cost must be split across consumers on a defensible basis.  
• Cloud billing data experience — AWS Cost and Usage Report, Azure cost exports, GCP billing export — and ideally familiarity with the FOCUS specification.  

Ramp and delivery expectation  

• Within two weeks: profiling assigned sources and producing design documentation the engineers can build from.  
• By end of December: silver-through-gold design complete for assigned integrations, and the application identifier triage advanced through the material end of the spend rank.