Hacker News new | ask | show | jobs
Show HN: LLM-powered data extraction from messy spreadsheets (github.com)
1 points by mehdig10 167 days ago
Uses an LLM to identify where tables actually start/end in Excel/CSV files, then extracts clean, typed data.

Handles all the messy formatting (currency, percentages, different number formats) and streams efficiently for large files.

Works with OpenAI, DeepSeek, or any compatible API.