NOTICE

LeanAI — Project-Aware Local AI Coding System
Copyright (C) 2025-2026 Gowri Shankar (@gowrishankar-infra)

This product was created by Gowri Shankar (Hyderabad, India).
GitHub: https://github.com/gowrishankar-infra/leanai

ORIGINAL AUTHOR ATTRIBUTION REQUIRED
=====================================
This notice must be preserved in all copies or substantial portions
of the software, including forks, derivatives, and redistributions.

If you use LeanAI or any substantial portion of its code in your
project, you must:
  1. Credit "Gowri Shankar" as the original author
  2. Link to the original repository
  3. Keep this NOTICE file intact

PROJECT HISTORY
===============
- Architecture designed and directed by Gowri Shankar
- Development started: April 2025
- 45 integrated technologies, 600+ tests, 36,000+ lines, 4 models with intelligent routing, streaming output, AGAC auto-correction, DFSG few-shot grounding, CodeEcho, DualPipe, repetition safety, LEANAI_HOME/LEANAI_MODELS support
- Built on: Windows 11, i7-11800H, 32GB RAM, RTX 3050 Ti
- Models: Qwen2.5 Coder 7B, Qwen3 Coder 30B MoE, Qwen3.5 27B, Gemma 4 26B MoE

THIRD-PARTY COMPONENTS
======================
- llama-cpp-python (MIT License) — LLM inference
- ChromaDB (Apache 2.0) — vector database
- FastAPI (MIT License) — web server
- Qwen2.5 Coder (Apache 2.0) — language model
- Qwen3 Coder (Apache 2.0) — language model
- Qwen3.5 (Apache 2.0) — language model
- Gemma 4 (Apache 2.0) — language model
- sentence-transformers (Apache 2.0) — embeddings
- psutil (BSD License) — system RAM detection