From my experience Machine learning (ML) workloads on Kubernetes (in the Cloud) offer unparalleled flexibility and scalability which is great but also can lead to the higher cloud spend. In this session, In this session, I'll be addressing the hidden financial and technical challenges of running ML on Kubernetes, with a focus on cost optimization which is a big concern in modern MLOps environments. Kubernetes has powerful capabilities, but can become inefficient without good management, especially when it comes to resource allocation. I will go over common issues like overprovisioning, where allocating more resources than needed leads to inflated costs, and will present practical strategies to "fine-tune" cloud resource usage for more efficient ML operations. In my opinion one of the most overlooked aspects of ML workloads is the cost of data movement—transferring large datasets between cloud storage and compute nodes, especially during frequent model retraining. In this session, I'll share my experience and actionable techniques to optimize data flow, reduce transfer costs, and accelerate model training cycles.

My name is Natalie. I'm a Staff Cloud Engineer.

If I have time, I also enjoy working on building and automating various tools that help our development team be more productive and happy. What motivates me at work is the fast pace, team orientation, and creative environment, always new challenges.

I'm passionate about helping make infrastructure more accessible. I love solving hard problems and all things containers. I enjoy being able to help engineers learn new ways to solve problems they are facing. I consider myself one of the not many engineers out there who worked on the highest number of CI/CD systems (because I enjoy making things better in the Software Release Processes at any company I work for)

Occasionally I blog or speak at conferences. I'm a technical mentor. My mentoring expertise is in the following topics: career development, self-improvement, interviewing, collaborating, communication, and soft skills - I have experience mentoring on various tech topics related to Python programming language or Infrastructure/DevOps.

In my spare time, I hike or camp with my aussiedoodles Chai and June and stand-up paddle (SUP) across the Bay.