“Within a month of our migration from Kubeflow to Neu.ro, we increased the ML experiments we could run by 3x.”
“The world is exploding with cameras.” – Yashar Behzadi, CEO Synthesis AI
Current State of Global Computer Vision1
Driven by deep learning, the global computer vision market is expected to reach $33.5 billion by 2025, according to market research firm Omdia.
To date, computer vision has been expensive and hard to scale as it relies heavily on supervised learning that requires humans in the loop to label key image attributes. Besides the cost and time required for manual labeling, there are also significant ethical and privacy issues of using of real-world data.
Further, synthetic data often provides a higher quality result than real images. All of these issues are effectively solved by Synthesis AI’s synthetic data technology. By combining CGI technologies with novel generative AI models, their simple API enables the programmatic generation of millions of images with pixel-perfect labels.
The Business Challenge
Synthesis AI understands that scale is critical to the success of its customers.
“The name of the game in AI is scale. We’re looking to create increasingly more data with increasing diversity.” – Yashar Behzadi
With individual client demands expected to exceed hundreds of millions of synthetic data images per month, Synthesis AI sought to build a robust and scalable infrastructure from day one.
Like many AI-focused companies, Synthesis AI initially assigned their internal ML engineering team the task of building and maintaining their MLOps infrastructure (including coordination and management of on-prem and cloud compute resources, data, models, pipelines and workflows). For this purpose, the team chose Kubeflow, a popular open-source ML development platform as the foundation upon which they would build their ML development lifecycle.
After 6 months of building and re-building on Kubeflow, the team realized that they were spending as much time on MLOps as they were on ML. Their pipelines required constant maintenance, they needed to manually integrate and update every tool they sought to use, and their computation resources required constant attention and maintenance. They realized that Kubeflow itself is not a scalable MLOps solution.
Furthermore, Synthesis AI was limited by their cloud provider, facing allocation and infrastructure maintenance issues that severely complicated their model training process.
Synthesis AI needed solutions for both MLOps and cloud computing that allowed their ML Engineers to focus on the models.
First, Neu.ro migrated the company’s entire ML workflow from Kubeflow to Neu.ro Platform. Second, Neu.ro engaged AWS, the global leader in cloud-based AI/ML infrastructure, to provide a long-term solution to Synthesis AI’s computational resource requirements, ensuring both scalability and availability. These migrations included:
- Deployment of Neu.ro cluster in the team’s existing infrastructure in the legacy environment
- Migration of the ML team from Kubeflow to Neu.ro
- Set up of infrastructure on AWS, including allocation of required computational quotas
- Deployment of the Neu.ro cluster on new AWS infrastructure
- Migration of data and computations from legacy cloud to AWS
As a result, Synthesis AI’s ML productivity tripled in the first month, while saving over $100,000 in cloud computing costs.
But, most importantly, Neu.ro unblocked Synthesis AI to scale its synthetic data platform.