Speakers
Description
As NASA science workloads expand across cloud platforms, maintaining financial visibility and controlling operational costs has become increasingly important. Without continuous monitoring, budget variances and inefficient resource utilization are often identified only after billing cycles close, limiting opportunities for corrective action and making cost containment of cloud service use significantly harder.We present an AI-assisted Financial Operations (FinOps) framework that provides continuous cost visibility, automated reporting, anomaly detection, and infrastructure optimization across the HSDcloud environment, without requiring a dedicated FinOps person.
The framework also measures the operational costs of AI services alongside traditional cloud infrastructure, enabling direct comparison between AI investment and realized infrastructure savings. The FinOps observability layer itself adds less than $1.50 per month in AWS overhead and is built entirely on existing infrastructure with no new servers or services required. In practice the use of this software is typically recovered quickly as even a single optimization recommendation is likely of greater value than the cost to run this software. This work demonstrates how AI-assisted FinOps improves fiscal stewardship, operational transparency, and long-term sustainability for NASA scientific cloud infrastructure while providing a practical model for cloud financial governance across research organizations.