Scale requirements
We're continuously increasing our scale constraints to support larger traffic loads.
Our published limits are intentionally conservative — in practice, we routinely support 2k+ QPS, up to ~1B items, and billions of events. If you have higher requirements, please reach out.
How we ingest and process your data at scale
Shaped has a real-time, durable feature-store that ingests, and transforms your data for training and serving.
How we train and encode data continuously
Shaped continuously encodes your data with fresh multi-modal embeddings and trains your retrieval and ranking models with best in class MLOps.
How we serve query results fast
Shaped has a serverless real-time serving system that scales with your requests based on latency, request volume and pod memory.
Scale limits
Shaped’s cloud API supports the following scale limits for each tenant:
Dimension | Limit |
|---|---|
| Unique users | 50 million |
| Unique items | 1 billion |
| Unique events | 1+ billion |
| Personal filters | 50 million |
| Requests per second | 2,000+ |
| Train frequency | 2 hours |
| Event and filter ingestion | < 30 seconds |
| User and item catalog ingestion | 10 minute |
Please get in touch if you have specific performance or scale constraints that you want us to meet: Schedule a call