AWS re:Invent is a learning conference hosted by AWS for the global cloud-computing community.
December 2nd, 2024 - December 6th, 2024
Las Vegas, Nevada
320
Read More
Deploy AI/ML Development Tooling at Scale: Key Takeaways from Our June Webinar
Marc Paquette
on July 18th, 2024
Updated on October 16th, 2024
Generative AI has found a useful place in software development, the advantages are clear. But how can you effectively deploy these tools within your own development organization? Our recent webinar dives into the practical steps to implement a cost-effective solution that meets your developers' AI/ML needs without overspending on GPU resources.
Coder on Kubernetes is a powerful combination with many advantages over conventional methods for deploying AI/ML for developers: it saves time, reduces costs, reduces security risks, and gives developers a positive, productive experience.
How do we know this about AI/ML in development environments? We’ve helped many medium and large developer organizations not just in the high-tech industries you’d expect, but in other vertical markets like insurance, healthcare, oil and gas, utilities, and transportation.
Building on insights from our May webinar on GenAI in software development, Tim Quinlan, Coder’s Senior Technical Marketer, gives a technical discussion of implementing AI/ML tooling for your developers.
A common AI/ML deployment method
A common way to deploy AI/ML for software development is to use physical, on-prem GPUs and monolithic VMs. This setup gives quick initial gains, to be sure, but an organization also has to manage security, predictability, and hope to find sufficiently skilled human resources. And this method is resource intensive: developers often end up waiting for access to a limited pool of GPUs or the same GPUs sit idle outside of business hours.
Coder on Kubernetes brings automatic scalability
Coder on Kubernetes scales GPU resources automatically, so your organization pays only for what it uses.
Coder Enterprise’s RBAC controls give the flexibility of providing tiered workspaces to fit the project. Coder assigns appropriate GPU resources whether developers are working on GPU-intensive projects or more modest efforts.
Deploying AI/ML in development environments
In a demo of Coder on Kubernetes, Tim starts with Kubernetes configuration, including GPU provisioning, nodes, and taints and labels. Then he covers Coder setup, including node selection and toleration, requests and limits, and RBAC.
Coder uses templates to provision developer workspaces. Tim shows the template that manages all these features to automatically configure an environment that’s tailored to the developer’s needs.
Learn more
Coder on Kubernetes is a cost-effective way to deploy scalable AI/ML in development environments while giving good developer experience, security management, and flexibility.
If you want to watch the full webinar, dive deeper by registering here. You can also check out our repo with example Coder templates for AI/ML deployment.
Join our upcoming webinar: a large global financial services company will share their Coder journey in the next webinar on June 25th, 2024.
Subscribe to our Newsletter
Want to stay up to date on all things Coder? Subscribe to our monthly newsletter and be the first to know when we release new things!