CSV Cleaner & Transformer — Wrangle in Plain English
A live, gated demo: load a CSV, then describe a cleanup in plain English — trim, standardize, split columns, deduplicate, filter. The AI writes a safe read-only transformation, shows a preview, and lets you download the cleaned file. LLM-agnostic; this instance uses Gemini.
Overview
CSV Cleaner & Transformer lets anyone reshape and clean data by describing the change in plain English. Instructions like "trim whitespace, standardize dates to ISO, split full name into first and last, dedupe by email" are translated into a single read-only SQL transformation, executed against an in-memory SQLite engine, previewed before/after, and downloadable as a cleaned CSV. Only read-only SELECT transformations run, and the generated SQL is always shown for transparency. LLM-agnostic — runs on a local model (Ollama / Llama) or any cloud LLM; this deployment uses Gemini. Data is processed in-memory and expires automatically.
Problem Solved
Data wrangling is repetitive and usually requires SQL or a script for every small change. This demo shows how natural language plus a safe execution layer lets non-technical users clean and reshape data in seconds — with the generated SQL exposed so the result can be trusted and reused.
Capabilities
- Plain-English → SQL data transformations
- Trim, case-normalize, cast, split, replace, filter, deduplicate
- Read-only guardrail (SELECT-only, single statement)
- Before/after preview with transparent generated SQL
- Download the cleaned CSV
- CSV upload or built-in sample datasets
- In-memory processing, rate limiting, gated access