Senior Data Analyst Public Cloud LRP Model Automation

Senior Data Analyst Public Cloud LRP Model Automation

Hybrid role. Tuesday-Thursday onsite. PST hours preferred.



About the Role: We are seeking a Senior Data Analyst to lead the end-to-end conversion and automation of our Public Cloud Cost & Usage Long Range Planning (LRP) model. The current model is an Excel-based, driver-driven framework that projects multi-year public cloud spend across compute, storage, networking, and managed services. This role will migrate that model into a maintainable, production-grade Jupyter Notebooks environment complete with automated source-system pipelines, versioned model snapshots, and a Monte Carlo simulation engine for probabilistic forecasting.



This is a high-visibility, high-ownership role that combines financial modelling expertise with data engineering capability. The successful candidate will need to understand the model's existing business logic deeply, build robust pipelines to replace manual data inputs, and deliver an architecture that makes every version of the model permanently auditable enabling precise variance analysis between actuals and forecasts, and between successive forecast versions.



Key Responsibilities

1. Excel LRP Model Analysis & Migration

? Conduct a full structural audit of the existing Excel LRP model: document every driver, input assumption, formula chain, interdependency, and output metric.

? Map the driver hierarchy identifying primary cost drivers (e.g. workload growth, instance mix, Unit Price per Region, SKUs per Region per each hyperscaler) and their upstream input sources.

? Re-engineer the entire model in Python within a modular Jupyter Notebooks architecture, preserving exact calculation fidelity while eliminating manual steps.

? Validate migrated outputs against Excel results line-by-line before decommissioning the spreadsheet workflow.



2. Source System Pipeline Engineering

? Design and build automated data ingestion pipelines for all model input feeds, including:

- Cost & Usage APIs to the homegrown FinOps Public Cloud Control Tower which hosts actual Cost & Usage of AWS, Azure, and Google Cloud Platform

- Migration demand Sales input of Renewal activities which triggers migration to public cloud and the relevant workload forecasts

- Rate card feeds Net Effective Cost per each Hyperscaler Region per SKU

? Implement data validation, reconciliation checks, and alerting so pipeline failures surface immediately rather than silently corrupting model inputs.

? Store versioned, timestamped snapshots of all raw input data to support full auditability of every model run.

? Abstract pipeline connections behind a clean configuration layer so source system changes require minimal model rework.



3. Driver-Based Calculation Engine

? Build a parameterised calculation engine that accepts driver assumptions as structured inputs (YAML / JSON config or a dedicated assumptions notebook) and propagates them through the model deterministically.

? Implement the full driver tree: usage growth rates

? Ensure the engine supports both a base case run and a stochastic simulation run from the same codebase.

? Produce standard LRP output tables: monthly/quarterly/annual spend by each hyperscaler region and by Purpose.



4. Model Versioning & Variance Analysis Framework

? Implement a versioning system that saves a complete, immutable snapshot of every model run capturing inputs, assumptions, driver values, and full output tables with a timestamp and named version tag.

? Build a variance analysis module that can compare any two saved versions and decompose the difference into named drivers, including:

- Actual vs. Forecast (budget vs. actuals bridge)

- Current Forecast vs. Prior Forecast (re-forecast bridge, explaining what changed and why)

- Scenario vs. Scenario (e.g. Base vs. Upside vs. Downside)

? Generate automated variance commentary structured output that attributes movements to specific drivers (e.g. 'RI coverage rate change +$2.1M; workload growth assumption increase +$3.4M').

? Expose a simple notebook interface that allows Finance analysts to select any two versions and instantly render the variance bridge without writing code.



5. Monte Carlo Simulation Engine

? Design and implement a Monte Carlo simulation layer that treats key model drivers as probability distributions rather than point estimates.

? Calibrate distributions (normal, log-normal, triangular, uniform, or empirical) against historical actuals and forward-looking business intelligence.

? Run simulations (target: 10,000+ iterations) to produce probabilistic output distributions for total cloud spend and major cost categories.

? Generate P10 / P50 / P90 confidence interval outputs for executive scenario planning and risk-adjusted budget setting.

? Build sensitivity / tornado analysis to rank drivers by their contribution to forecast variance.



6. Reporting, Visualisation & Stakeholder Enablement

? Build interactive output dashboards within notebooks covering: waterfall variance bridges, scenario fan charts, driver sensitivity charts, and spend-by-category treemaps.

? Automate recurring model refresh workflows so Finance team members can re-run the full model with a single command.

? Author comprehensive documentation: model methodology, driver definitions, pipeline architecture, versioning guide, and assumption dictionary.

? Present model outputs, variance bridges, and scenario analyses to Finance leadership and C-suite stakeholders.



Required Skills & Experience

Jupyter Notebooks Financial Modelling Mandatory

? Demonstrated, hands-on experience building production financial models directly in Jupyter Notebooks. Portfolio / GitHub links required analysis-only notebooks do not qualify.

? Proven track record implementing Monte Carlo simulations in Python: distribution fitting, correlated variable sampling, simulation loop design, and P10/P50/P90 output interpretation.

? Proficiency in core Python modelling stack: NumPy, Pandas, SciPy (stats), and at least one Monte Carlo framework (PyMC, NumPyro, custom simulation engine, or equivalent).

? Experience with Plotly, Matplotlib, or Bokeh for financial chart types: waterfall bridges, fan charts, tornado charts.



Excel LRP / Driver-Based Model Experience Mandatory

? Direct experience working with or converting complex Excel-based LRP, 3-statement, or driver-based financial models (multi-tab, formula-intensive, 3-5 year horizon).

? Ability to reverse-engineer Excel model logic tracing precedents, documenting assumptions, and translating formula chains into equivalent Python with verified numerical accuracy.

? Understanding of driver-based modelling methodology: separating volume drivers from price/rate drivers, building assumption sensitivity tables, structuring base/upside/downside scenarios.



Source System Pipeline Engineering

? Ability to build lightweight ETL/ELT pipelines in Python: API authentication, pagination, schema normalisation, error handling, and incremental refresh logic.

? SQL proficiency for data extraction from billing databases, data warehouses

? Experience with data pipeline orchestration tools (Airflow, Prefect, dbt, or notebook scheduling) is advantageous.



Model Versioning & Variance Analysis

? Experience designing model version control beyond Git structured snapshot storage of inputs, assumptions, and outputs to enable point-in-time model reconstruction.

? Familiarity with variance bridge / waterfall decomposition methodologies used in FP&A (price-volume-mix, driver attribution).

? Comfort building automated commentary or structured output that explains numerical movements in business terms.



Qualifications

? Bachelor's degree in Finance, Economics, Mathematics, Computer Science, or a quantitative discipline. CFA, CIMA, CPA, or AFP/FP&A certification is a plus.

? 5+ years in financial modelling, FP&A, or cloud economics roles.

? 3+ years of hands-on Python / Jupyter financial modelling (not data science alone).

? Demonstrable Git proficiency; experience with code review in a team modelling context is preferred.