FinOps Engineering
Cost Engineering as
Architecture Discipline
FinOps is not bill reporting. It is the engineering discipline of making cost drivers visible, choosing the right infrastructure envelope, and building systems that teams can operate without waste.
Case Study
AWS Cost Dashboard — 78% Reduction
Two deployment architectures for the same FinOps dashboard. The cost difference came from choosing the right infrastructure envelope, not from cutting scope.
ECS Fargate + ALB
ECS Fargate with ALB provided enterprise-grade delivery. Correct architecture for a high-traffic production workload — but the dashboard's actual traffic pattern did not justify per-task Fargate pricing or a dedicated load balancer.
View V1 Architecture →EC2 t4g.nano + Caddy
Same dashboard, same Lambda + Athena data pipeline, same GitLab CI lifecycle. Swapped the compute and serving layer to match the workload's actual profile: low-traffic, static-heavy, no scaling requirement.
View V2 Architecture →Decision journey
V1 — Enterprise architecture
Fargate + ALB at $27/mo. Correct for an enterprise deployment, but overkill for the workload profile.
Analysis
Fargate per-task pricing was wrong for this workload. ALB added fixed cost with no scaling benefit. The cost envelope did not match the demand pattern.
V2 — Cost-optimized redesign
EC2 t4g.nano + Caddy at $6/mo. Same dashboard, same Lambda + Athena data pipeline, 78% lower cost.
Outcome
The reduction came from choosing a different infrastructure envelope — not from turning things off, cutting features, or chasing reserved instances.
Approach
How I Approach FinOps
These are the principles that shaped the cost dashboard redesign and run through the rest of my infrastructure work.
Start with architecture, not dashboards
Cost problems are architecture problems. A dashboard that shows overspend on a poorly designed system is reporting failure, not preventing it.
Make cost drivers visible to engineering teams
Cost decisions belong where infrastructure decisions are made. Finance reports matter, but engineers need to see what their design choices cost in real time.
Choose the right infrastructure envelope
Right-sizing is not about picking smaller instances. It is about choosing the compute model that fits the workload profile — then validating it.
Build sustainable practices, not one-time cuts
A one-off cost reduction that nobody understands six months later is not FinOps. The system has to be supportable by the next engineer.
Stack Knowledge
AWS FinOps Data Pipeline
The AWS FinOps stack follows a clear data flow: billing data is generated, transformed, queried, and visualized. Understanding this pipeline end-to-end is what separates cost reporting from cost engineering.
CUR / CUR 2.0
Billing data source
AWS Glue
ETL and data preparation
Amazon Athena
Serverless SQL analysis
Amazon QuickSight
Visualization and sharing
Cloud Intelligence Dashboards
Pre-built FinOps views (CID, CUDOS)
The case study dashboard uses Lambda + Athena for the query layer and a lightweight frontend for visualization. The V1-to-V2 cost reduction happened in the serving layer, not the data pipeline — proving that FinOps optimization targets the right component, not every component.
Beyond AWS
FinOps Mindset Across the Homelab
The same cost discipline runs through the entire homelab. Self-hosting on Proxmox instead of paying for cloud hosting for everything is itself a FinOps decision: trading operational effort for cost control, with full visibility into what that tradeoff costs.
The FinOps mindset is architectural, not tool-specific. It applies whether the workload runs on AWS, Azure, or bare metal in a home rack.
AWS Cost Dashboard
78% cost reduction through architecture-led redesign
Self-hosted infrastructure
Proxmox + Kubernetes replacing cloud hosting where it makes sense
CUR data pipeline
Reduced Lambda + Athena invocations by tuning refresh to CUR update cadence (1-3x/day)
Portfolio site
Docker + Caddy on Proxmox — zero external hosting cost
FinOps Foundation Member
Since 2025
FinOps is a growing part of my platform engineering work. The Foundation membership reflects commitment to the practice — the case study and architecture work above reflect how I apply it.
Interested in this work?
Open to senior cloud, platform, and FinOps engineering roles where cost discipline is part of the architecture conversation.