Project Case Study

Back to Projects

FloatChat

Optimizing querying and storage for raw NetCDF datasets for scientific analysis.

Architecture

Hybrid architecture with SQLite indexing for geospatial lookups and an ETL pipeline converting datasets into partitioned Parquet files.

System Diagrams

ETL and Hybrid Retrieval Architecture

Tech Stack

SQLite
ChromaDB
RAG

Outcomes

Achieved 10-100x faster query performance and reduced storage footprint by 60-80% via ChromaDB compression.

Source Code

https://github.com/HimeshRaj77/FloatChat.git