April 17, 2023 marked a couple of months into my new role as an engineering manager. I was still getting my feet wet with the challenges and responsibilities that came with it, especially in navigating the complex landscape of cloud costs and infrastructure management. The tech world was in full swing, with AI/ML tools like ChatGPT and StableLM pushing boundaries, and FinOps becoming a mainstream buzzword. It felt like I was in the middle of a whirlwind of change, yet I found myself wrestling with the same old problem—how to balance the need for cutting-edge technology with keeping costs under control.

One of my first big projects as an engineering manager involved optimizing our infrastructure cost management using tools and practices from FinOps. Our team had grown significantly over the last few years, and with it came a growing concern about cloud spend. We were using platforms like AWS and GCP, which, while powerful, also come at a steep price when not managed well.

The transition to FinOps meant that I was no longer just focused on building great products; now, I had to ensure every dollar spent was justified by business value. It was a shift from the traditional “let’s just throw more resources at it” mentality to one of intentional and strategic allocation.

One of our key initiatives involved implementing cost-aware deployment strategies using tools like Cost Explorer and Budgets in AWS. We were leveraging these tools to set up alerts for unusual spending patterns and to identify opportunities for optimization. This required a deep dive into understanding our usage patterns, which meant we had to collect detailed metrics on everything from CPU utilization to API calls.

I remember the frustration of debugging these complex configurations. It was like trying to untangle a Gordian knot—every service was intertwined with multiple layers of dependencies and billing structures. At times, I felt like a hamster running in circles on a wheel, but there were moments when it all started to come together. For instance, we managed to reduce our monthly spend by 20% through a series of targeted optimizations.

Another challenge was aligning the development team with these new FinOps practices. There’s always a learning curve when introducing new methodologies and tools. Some developers were resistant at first, thinking they had more pressing concerns like shipping features on time. But once we started showing tangible results—like cost savings that could be reinvested in projects—the skepticism turned to understanding.

During this journey, I found myself reflecting on the lessons from past projects and technologies. For example, while WebAssembly was promising for running compiled code on servers, its adoption wasn’t as widespread as some hyped it to be. We decided to keep an eye on it but focus more on improving our existing infrastructure with proven techniques.

The era of AI/ML tools like StableLM added another layer of complexity. While these tools opened up exciting possibilities, we had to ensure they didn’t become a cost driver. I recall one particularly intense debate about whether to use pre-trained models or build custom solutions in-house. In the end, we opted for a hybrid approach, using off-the-shelf models where feasible but customizing them when necessary.

As I look back at this period, I realize that managing FinOps is not just about saving money; it’s about making strategic decisions and fostering a culture of mindful resource management. The tech world may be full of new tools and innovations, but the core principles of good engineering—like understanding your systems deeply and making informed choices—are timeless.

In the end, the journey of learning to balance cutting-edge technology with cost-conscious practices has been a humbling yet rewarding experience. It’s taught me that sometimes, the most valuable lessons come from the simple act of doing something small and consistent over time, rather than trying to reinvent everything overnight.

---

That was my thought process as I navigated those months. It felt like I was walking a tightrope between innovation and pragmatism, all while keeping an eye on the ever-changing tech landscape.