Cloud Spend Analytics MVP with Forecasting and Cost Spike Detection

Overview

We need a developer to create an MVP dashboard for analyzing cloud infrastructure spending.

The platform should help users upload cloud cost data, understand spending patterns, identify unusual increases, and generate a basic forecast for future costs.

This is intended to be an internal analytics tool, not a full enterprise SaaS product. The main focus is clean functionality, useful insights, and a codebase that can be extended later.

Core Objective

Build a working web application that can process historical cloud cost data and present it in a clear dashboard with forecasting and anomaly detection features.

Functional Requirements

1. Data Import

The system should allow users to upload cloud billing data in CSV format.

The CSV may include fields such as:

Date
Service name
Region
Cost
Usage type
Account or project name

The application should validate the file, process the data, and store it in a database.

2. Cost Dashboard

The dashboard should display:

Total spend over time
Daily and monthly cost trends
Breakdown by service
Highest-cost services
Cost changes compared to previous periods

Charts should be simple, readable, and useful.

3. Forecasting

The application should generate a basic cost forecast for the next 7–30 days using historical data.

A simple statistical or machine learning approach is acceptable. The prediction does not need to be extremely advanced, but it should be explainable and reasonably useful.

4. Anomaly Detection

The system should detect unusual cost spikes and highlight them in the dashboard.

For example:

A service suddenly becomes much more expensive
Daily spending is significantly higher than normal
A specific category shows abnormal growth

Each anomaly should include a short explanation.

5. Backend API

The backend should provide endpoints for:

Uploading data
Retrieving dashboard metrics
Getting forecast results
Listing detected anomalies

The code should be clean and modular.

Preferred Stack

Python
FastAPI
Pandas
Scikit-learn, Prophet, or another simple forecasting method
PostgreSQL or SQLite
React for frontend
Docker would be helpful but is not mandatory

Acceptance Criteria

The project will be considered complete when:

CSV cost data can be uploaded successfully
The dashboard shows cost trends and service breakdowns
The system produces a 7–30 day forecast
The system detects and displays unusual cost spikes
The application can run locally with clear instructions
The code is organized and suitable for future improvements

Additional Notes

Please include a small sample dataset or explain the expected CSV format clearly. Also include a short technical note describing how the forecasting and anomaly detection logic works.