This repository contains a proprietary Cloud Hosting Forecasting Engine designed to provide 100% visibility into multi-cloud hosting costs (AWS & GCP). In a high-scale environment like Reddit, cloud spend is the primary driver of Cost of Revenue (COR).
This framework does not just track spend; it enforces Financial Governance, automates Variance Analysis, and provides the data-backed leverage required for Strategic Contract Negotiations (AWS EDP / GCP CUD).
.png)
Figure 1: 90-Day forward-looking projection identifying strategic scaling risks and contract commitment windows.
Using seasonal time-series modeling (Prophet), this engine identifies the "Drivers of Change" within infrastructure spend. It accounts for weekly traffic spikes and organic growth, reducing budget variance to <2%.
The engine includes a data-governance layer that normalizes raw CUR data. It ensures 100% of spend is allocated to specific Product LoBs (e.g., AIResearch, Ads, CorePlatform), allowing for precise Unit Economics (Cost per Daily Active User).
Through scenario_config.json, the engine conducts rapid impact analysis for:
- Product Launches: Projected infrastructure burn for new feature deployments.
- Scaling Events: Impact of 1.2x - 1.5x user growth on monthly gross margins.
- Contract ROI: Pre-calculating the break-even point for multi-year cloud commitments.
| Phase | Component | Strategic Output |
|---|---|---|
| Ingestion | scripts/preprocess.py |
Normalizes AWS/GCP data; enforces tagging governance. |
| Modeling | notebooks/forecasting.py |
Seasonal spend projection & trend decomposition. |
| Strategy | scenarios/scenario_config.json |
Scenario analysis for product scaling & contract ROI. |
| Visibility | dashboards/variance_dashboard.py |
Automated Forecast vs. Actuals reconciliation. |
| Executive | reports/variance_report.md |
Actionable insights for the CFO & VP of Engineering. |
Figure 2: Automated Forecast vs. Actuals reconciliation used for monthly close and engineering accountability.
- Time-Series Engine: Utilizes the Prophet library to handle non-linear trends with yearly, weekly, and daily seasonality + holiday effects.
- Data Normalization: The
preprocess.pyscript acts as a governance gate, cleaning raw Billing Exports (CUR) and mapping them to internal cost centers. - Scalability: Designed to handle multi-million row billing files common in Enterprise AWS/GCP environments.
To protect the proprietary FinOps logic and internal cost-center mappings, functional source code is restricted. This repository serves as an Architectural Showcase of my ability to build and lead Technology FP&A functions at scale.
For a full demonstration or architectural deep-dive:
- Contact: Jeya Prakash I
- Role: Sr. Technology FP&A Manager Candidate
- Specialization: Cloud FinOps, Strategic Planning, Contract Negotiation.