September 19, 2022. The smell of autumn is in the air and with it comes the annual tech industry introspection—looking back at what we accomplished, pondering where to go next, and often, being reminded that we’re not doing enough.

Today, I can’t help but reflect on something that’s been niggling at me for a while now: FinOps. For those who might be new to the term, FinOps (Financial Operations) is the practice of aligning technology spending with business objectives by treating cloud spend like a product or service and integrating it with development and operations teams.

In my previous post, I delved into how platform engineering has become mainstream—teams like mine are no longer just about building reliable systems; we’re also tasked with ensuring those systems are cost-effective. This shift is part of the broader FinOps movement that’s gaining traction as cloud costs balloon.

Let me share a recent experience that brought this home in a very personal way. Over the summer, I was working on optimizing our application for the latest AI/ML workloads using PyTorch and TensorFlow. As we scaled up the infrastructure to support these models, my eye kept being drawn back to the cloud cost metrics. The spike in costs due to model training was concerning, but it also made me think about the broader implications.

One day, I was deep into a debug session when I noticed something alarming: our AI/ML team was spending over $10k per month just on data transfer between S3 and the compute instances. This wasn’t an isolated issue; we had multiple teams with similar problems. The irony wasn’t lost on me that while we were obsessed with writing efficient code, we hadn’t given as much thought to optimizing our cloud spend.

This realization hit hard. We needed a better way to manage costs. So, I started exploring FinOps tools and practices. One thing led to another, and before I knew it, I was diving into AWS Cost Explorer, Azure Budgets, and Google Cloud’s Pricing Calculator API. It was like stepping into a whole new realm of ops work that I hadn’t fully appreciated.

One particularly frustrating day, I found myself in an argument with my team about whether we should spin up more EC2 instances or use Spot Instances for training jobs. The debate wasn’t just about technical trade-offs but also financial ones. We had to weigh the benefits of having predictable resources against the risk of unpredictable costs. It was a reminder that every decision we make in platform engineering has financial implications.

I remember feeling overwhelmed by the sheer volume of data and metrics these tools generated. AWS Trusted Advisor, Azure Cost Management + Billing—everywhere I looked, there were recommendations, alerts, and metrics. The CNCF landscape seemed to be expanding even further into this space with new tools like OpenFaaS and Kubeflow integrating more closely with cloud providers.

But amidst all the complexity, a simple realization dawned on me: FinOps is not just about saving money; it’s about making better-informed decisions that balance cost with performance. It forces us to think about our tech stack in a different way—more like a financial portfolio than a series of development tools.

As I write this, I’m still grappling with the best practices for implementing FinOps in my team. There’s no one-size-fits-all solution, and every organization will need to find its own path. What I do know is that we can’t afford to ignore these costs any longer. The tech industry has a reputation for reckless spending, but as platform engineers, it’s time we step up and take ownership of our financial impact.

So, here’s my advice: if you’re like me and haven’t fully embraced FinOps yet, start by looking at your own team’s cloud spend. Use tools like AWS Cost Explorer or Azure Budgets to get a baseline understanding. Then, begin asking questions about why certain services are being used and how they can be optimized.

Remember, the goal isn’t just to cut costs but to ensure that our tech decisions align with business objectives while staying within budget constraints. It’s a new discipline that requires a different mindset—one that balances development velocity with financial responsibility.

In the end, FinOps is not about saving money at any cost; it’s about making smarter investments in technology that drive real value for your organization. And as we move forward, I’m excited to see how this practice evolves and becomes more integrated into our daily operations.

Stay tuned for more updates as I navigate this new frontier!