ACO — AT&T Cloud Optimizer

NDA

2021 · AT&T (Chief Data Office)

Internal tool that helps AT&T org leaders see, attribute, and right-size their Azure cloud spend.

Methodology, research, and iteration details are restricted under NDA. The screens below are shareable; all on-screen names and figures are illustrative sample data, not real AT&T personnel or spend.

ACO is an enterprise Azure cost-optimization dashboard built for AT&T's Chief Data Office. It models the org as a 4-level hierarchy (Organizations → Teams → Applications → Resources) so VPs and app owners can drill from a top-line cost number to the exact VM/VMSS that's overprovisioned.

The problem

Before ACO, the Chief Data Office had no single, attributable view of Azure spend across the enterprise. Cloud cost was spread over hundreds of applications and dozens of org units, so VPs couldn't see what their teams were spending, app owners had no accountability for over-provisioned resources, and right-sizing was a manual, one-off analysis. Finance could see the total bill but had no model to allocate or reduce it.

Information architecture

All Organizations  (executive cards)
  └─ Organization
        ├─ Overview        apps · spend · resources · VM/VMSS split
        ├─ Teams           per team lead: 13-month cost trend + savings opportunity
        └─ Applications    table: app · MOTS ID · type · owner · in-year cost · savings
              └─ Application
                    ├─ Overview
                    └─ Subscriptions
                          └─ Resource
                                ├─ Details          SKU · CPU / RAM / IOPS / Throughput
                                ├─ Recommendations  alternative SKUs ranked by savings
                                ├─ Risk Level       conservative ↔ aggressive
                                └─ Metrics          allocation vs utilization over time

A parallel Organization Tree mirrors the reporting hierarchy (SVP → VP → team lead → app group → application), searchable by person or MOTS ID.

Process & prototyping

Consistent IA across every entity level: top-line KPI strip + 13-month cost-trend sparkline + savings-opportunity callout.
Action badges (red / yellow / N/A) surface recommendation counts at every row in the apps table.
Recommendation engine ranks alternative Azure SKUs by projected savings, with a tooltip explaining the 30-day-percentile basis.
Organization Tree view (search by user or MOTS ID) supports hierarchical drilldown by SVP → VP → Team Leader → App Group Leader → Application.
Per-resource detail page exposes a Risk Level slider (High ↔ Low) that biases recommendations toward conservative vs aggressive right-sizing.

Screens

Tap any screen to enlarge.

LandingThe entry point routes users by intent — as an application owner, as a leader, or straight into search.

All organizationsEach organization is a card with a 13-month cost-trend sparkline, year-to-date spend, and a monthly savings-opportunity figure.

Organization overviewAn org rolls up applications, spend, and resources, with a consumable-metrics strip (CPU/RAM/IOPS/Throughput) and a YTD cost breakdown.

TeamsDrilling into teams: every team lead gets the same cost-trend and savings card, so spend stays attributable down the org chart.

ApplicationsThe applications table ranks each app by in-year cost and savings opportunity; color-coded badges flag how many right-sizing recommendations are open.

Application detailAn application pairs ownership and mission-criticality with its cost trend and consumable-metrics profile.

ResourcesWithin an application, every Azure resource is listed with its SKU, monthly cost, utilization, and savings opportunity.

RecommendationsThe recommendation engine ranks alternative SKUs by projected savings; a Risk Level slider biases the suggestions from conservative to aggressive.

Allocation vs utilizationThe core right-sizing chart: the gap between what's provisioned and what's actually used over time is the savings.

Organization treeAn expandable tree mirrors the real reporting hierarchy (SVP → VP → team lead → app group → application) for top-down navigation.

SearchGlobal search resolves a person, MOTS ID, or application and surfaces each result's open savings opportunity.

Outcome & impact

ACO gave every level of the org a drill-down from a top-line cost number to the specific over-provisioned VM, with ranked right-sizing recommendations and a risk dial to balance savings against stability. The platform contributed to an 18% enterprise-wide cloud-cost reduction at AT&T.

Role & collaboration

Team

~6 people, same team across both ACO and SUD: 2 frontend engineers (Kshitij + 1 other; both also did UI design work), 1 tech lead, 1 scrum master, 2 data engineers handling the Oracle → MongoDB ETL pipeline. Kshitij wasn't titled 'lead' but in practice led the backend, DevOps, and cloud avenues.

My role

Owned the majority of the frontend implementation (the other FE contributed UI work too).
Took over the backend in Node.js with Apollo GraphQL.
Owned DevOps + Kubernetes, including the Azure Key Vault → K8s secrets integration.
Owned the entire cloud deployment pipeline.
Co-designed the UI alongside the other frontend engineer.

Tools

Azure (East US 2 region)VM / VMSS SKUs (Standard_B8ms, Standard_DS1_v2)AT&T MOTS ID systemaiOps sibling product (aiOpX)