GenAI Analytics & Reporting

Overview

A web-based analytics and reporting platform for Walmart's multicloud capacity management system, built to transform raw infrastructure data into actionable insights at high velocity. The application provides engineering and operations teams with on-demand capacity reports across Walmart's hybrid cloud spanning OpenStack, GCP, and Azure.

The platform was designed from the ground up as a GenAI-assisted development showcase — the entire application was built in a fraction of the time of a traditional development cycle by leveraging WIBEY (Walmart's AI coding assistant) for implementation, test generation, and deployment automation.

GenAI-Driven Development

This project served as a live demonstration of GenAI-augmented development velocity. The web-based analytics and reporting platform provides engineering and operations teams with on-demand capacity reports across Walmart's hybrid cloud spanning OpenStack, GCP, and Azure showing infrastructure capacity data for compute, storage and network assets for both realtime analysis and long range forecasting and planning.

Benefits of GenAI Reporting Platform

• Report implementation generated from natural language requirements
• Implementation velocity shown to be orders of magnitude faster than report implementation using PowerBI and Looker
• Custom coded application functionality provides significant user interaction and display functionality improvements over structured reporting platforms

The core of the platform is a pluggable report module architecture. Each report registers itself at startup with metadata describing its parameters, data sources, and output formats. The web UI dynamically renders parameter forms based on the report's configuration — adding a new report requires only implementing the report module interface.

Reports supported at launch:

• Capacity Request Automation — Analyzes automation rates for capacity requests with Jira enrichment, fiscal quarter filtering, and period-over-period comparison
• Cloud Compute AVD — Month-over-month tracking of cloud compute capacity across GCP, Azure, and OpenStack
• GPU Capacity Demand — GPU utilization and demand analysis by organization and cloud provider
• GPU Request Validation — Cross-references Jira GPU tickets against PostgreSQL capacity records to surface mismatches
• OpenStack Capacity — OpenStack-specific compute and storage capacity metrics
• Automated Reclaim — Identifies unused or underutilized capacity eligible for reclamation

Architecture

The application is built on a modern Python stack with a focus on developer velocity and operational simplicity. The backend is a FastAPI application exposing both a web UI and a REST API, deployed as a containerized service on WCNP (Walmart's Kubernetes platform).

• Backend: FastAPI + Uvicorn (Python 3.12), with async job execution for long-running report generation
• Frontend: Jinja2 server-side templates with HTMX for dynamic UI updates without a JavaScript framework
• Data Sources: Google BigQuery (Tekton quota and capacity metrics) + PostgreSQL (capacity request tracking)
• Data Processing: Pandas and NumPy for aggregation, trend analysis, and period-over-period comparisons
• Export: OpenPyXL for Excel report generation with formatted multi-sheet workbooks
• Auth: PingFederate SSO via Istio sidecar; group-based authorization (capacity_mgmt)
• Deployment: Docker multi-stage build, KITT CI/CD pipeline, WCNP with liveness/readiness probes