Success story.
Serverless AI Architecture
DeepIA wanted to scale its AI models without managing complex infrastructure, streamlining the launch of new vision and NLP services.
We take on any challenge
At Zenta there are no obstacles we cannot overcome.
Challenge
Design a serverless architecture capable of deploying and scaling AI models with minimal latency and variable costs based on usage, while maintaining enterprise security.
Solution
We implemented Cloud Run for inference APIs, Vertex AI Endpoints for model training and deployment, and Cloud Storage as a model repository. All orchestrated with Terraform and monitored with Cloud Monitoring.
Results from our management
50% cost reduction in inference.
Automatic scaling to 0 instances in off-peak.
Response time <150 ms by inference.
“"GCP has been a fundamental pillar for our serverless architecture thanks to its low costs, countless tools, and high efficiency."”
What they say about our work.
We have the backing of those who have trusted us.