Welcome to Yuki
Yuki is a data cloud optimization platform that helps organizations reduce Snowflake costs and improve performance - without any code changes or manual tuning. It works by analyzing metadata, which means setup is safe, lightweight, and requires zero engineering effort.
Yuki is officially available on the Snowflake Marketplace, allowing teams to connect in minutes and start optimizing immediately.
What Yuki Does
Modern data teams spend significant time and budget managing Snowflake compute. Yuki automates this optimization layer and integrates directly with Snowflake using metadata APIs - with no query interception and no access to your data.
It continuously:
Monitors query and warehouse metadata to understand workload behavior - including query history, execution time, and warehouse utilization
Detects inefficiencies in warehouse sizing, runtime, and credit consumption
Based on observed workload patterns, automatically adjusts warehouse sizes and routes queries to the most suitable compute resources - ensuring optimal performance and cost efficiency without user intervention
The result: lower spend, faster queries, and full visibility into compute usage.
Because Yuki operates purely on metadata, there’s:
No code modification
No change to user behavior or query paths
No data exfiltration risk
Deployment Types
Yuki can be deployed in two modes depending on your data residency and security requirements:
1. Full SaaS
Yuki hosts and manages the full optimization layer. You simply connect your Snowflake account using a service user and key-pair authentication. No infrastructure setup, no maintenance - you start optimizing within minutes.
2. Hybrid Deployment
For organizations that require processing within their own VPC or Business Critical Snowflake accounts. You deploy the Yuki Proxy inside your environment via:
Helm Chart
Terraform
In this setup, Yuki runs entirely inside your environment, forwarding only optimization metadata over HTTPS to the Yuki Engine.
Metadata-Only by Design
Yuki never reads or copies customer data - only system metadata such as:
Query IDs, timestamps, and runtime
Warehouse sizes and credit usage
Execution statistics
This ensures compliance with strict privacy standards (SOC 2 Type II, GDPR) while maintaining full visibility into performance and spend.
Security and Isolation
No access to data - Yuki only reads Snowflake metadata tables (
ACCOUNT_USAGE,INFORMATION_SCHEMA, etc.).Secure connection - authentication is done via Snowflake service user and key-pair.
Optional Hybrid mode - run Yuki Proxy inside your environment to keep all metadata local.
Next Steps
Now that you understand how Yuki works, follow these guides to complete setup and start optimizing:
Last updated