Inspiration Text-to-SQL is one of the most impactful and widely used tools in the data science and analytics ecosystem, enabling seamless natural language interaction with complex data.

What it does Ryoma is an intelligent agent that leverages state-of-the-art research to tackle the text-to-SQL problem, making data querying more intuitive and accessible.

How we built it We drew inspiration from leading benchmarks and projects, such as:

BIRD Benchmark: https://arxiv.org/abs/2505.19988

Snowflake ReFoRCE: https://github.com/Snowflake-Labs/ReFoRCE

Challenges we faced Scalability has been our biggest challenge—ensuring the system remains performant and reliable under large-scale, real-world workloads.

Accomplishments we’re proud of Fully Open Source – transparent and community-driven Secure by design – prioritizing data privacy and safety Easy to adopt – designed with simple integration in mind

What we learned Building a robust text-to-SQL agent requires balancing cutting-edge research with practical engineering, especially when targeting enterprise-level scalability and reliability.

What’s next for Ryoma We’re focused on evolving Ryoma with more scalable, resilient infrastructure and expanding its capabilities to support richer analytics workflows and multi-modal data queries.

Built With

Share this project:

Updates