Deploying Self-Hosted GenAI models for an up and coming Creative Technology Company

Client Background

A creative technology company sought to leverage advanced AI capabilities to enhance its digital content creation. To this end, it aimed to deploy self-hosted AI models, specifically Stable Diffusion, in the cloud, enabling artists and designers to generate high-quality visuals efficiently. The company needed a scalable and secure infrastructure that integrated seamlessly with its existing systems to achieve this.

Challenge

The primary challenge was to design a scalable, cost-effective cloud environment capable of hosting and managing AI workloads while ensuring security, scalability, and ease of use. The company required a solution that allowed its teams to utilize cutting-edge AI models without the complexity of managing the underlying infrastructure.

Solution

The team collaborated with BSC Analytics to develop a comprehensive cloud solution built on AWS EKS (Elastic Kubernetes Service). The solution focused on the following key areas:

• AWS EKS Environment Setup
  BSC Analytics engineers established a Kubernetes environment on AWS EKS, ensuring it was optimized for running AI workloads. The environment was configured to handle the demands of Stable Diffusion while maintaining high availability and security.

• Infrastructure as Code with Terraform
  The entire infrastructure was provisioned using Terraform, allowing for repeatable and consistent deployments. This approach streamlined the setup process and ensured the infrastructure adhered to best practices.

• Docker and Kubernetes Integration
  Docker images for Stable Diffusion and supporting applications were built and deployed to the EKS cluster. Helm charts were used to manage the deployments, simplifying scaling and updating the AI models.

• ArgoCD for Continuous Deployment
  ArgoCD was implemented to manage the continuous deployment of applications within the Kubernetes cluster. This allowed the company to automate the deployment process, ensuring that updates and new features could seamlessly roll out.

• iOS Mobile Application
  In addition to the cloud infrastructure, BSC Analytics also developed a mobile application for iOS. This allowed DreamWorks ' users to generate and interact with AI-generated content directly from their mobile devices, extending the platform's accessibility.

Results

The deployment of the BSC Analytics solution significantly transformed DreamWorks' AI capabilities:

• Enhanced Creativity and Efficiency
  Artists and designers leveraged Stable Diffusion to create high-quality visuals quickly and efficiently, enabling faster content production.

• Scalability
  The Kubernetes-based solution provided a scalable framework, allowing the company to handle varying workloads and expand its AI capabilities as needed.

• Seamless Integration
  The cloud environment integrated smoothly with the company’s existing infrastructure, enabling a hybrid setup that maximized both on-premise and cloud resources.

• User Accessibility
  The iOS mobile application expanded the platform's reach, allowing users to generate AI content.

Conclusion

This case study highlights BSC Analytics' expertise in deploying AI models in the cloud. By leveraging AWS EKS, Terraform, and Docker, BSC Analytics delivered a scalable and secure solution that empowered the company to push the boundaries of creativity. The addition of the iOS mobile application further enhanced the platform's accessibility, making AI-powered content creation available anytime, anywhere.