Dockerfile / README.md
kkthakur's picture
Deploy Local Hybrid Engine
b336134
|
Raw
History Blame Contribute Delete
782 Bytes
metadata
title: Zero-LLM Engine
emoji: 📊
colorFrom: blue
colorTo: indigo
sdk: docker
app_port: 7860
pinned: false

Zero-LLM Data Engine

A high-scale, production-ready backend engine for natural-language Excel/CSV/Parquet editing, built with Python and Polars.

Local Development

To run the engine locally:

cd zero-llm-engine
pip install -r requirements.txt
uvicorn main:app --host 0.0.0.0 --port 8000 --reload

Features

  1. Hybrid Parser: Fast regex/keyword path with fallback to offline local embedding/spelling parser, local Qwen GGUF model, or Gemini cloud API.
  2. Polars Execution: Safe and fast data frame operations with concurrency, low latency, and memory efficiency.
  3. MVCC Versioning: Safe read/write concurrency via versioned files.