Update and rename README.md to Sphinx

Sphinx

@@ -0,0 +1,50 @@
+# Team Sphinx | Argo-Agentic Intelligence 🌊
+
+## Problem Statement / Idea
+* **What is the problem?**
+Oceanographic telemetry from the Argo float network is often siloed in complex, technical formats, making it difficult for researchers to query data using natural language or perform high-speed spatial-semantic analysis.
+
+* **Why is it important?**
+Fast access to metrics like ocean heat content and oxygen anomalies is critical for climate adaptation. Current workflows lack the ability to bridge the gap between native language queries (Hindi/Bengali) and technical SQL databases.
+
+* **Who are the target users?**
+Marine data scientists, environmental policy-makers, and researchers requiring offline-first, high-accuracy scientific assistance for deep-ocean telemetry.
+
+## Proposed Solution
+* **What are you building?**
+We have built the **Agentic Ocean Intelligence Hyperpipeline**—a localized ML framework and Supabase-ready ingestion system that transforms raw telemetry into a RAG-augmented knowledge base.
+
+* **How does it solve the problem?**
+By fusing **AutoGluon** deterministic models for query routing and **LangChain** for local LLM orchestration, we automate the path from a researcher's natural language question to a precise SQL or vector-based answer.
+
+* **What makes your solution unique?**
+The system is **Sovereign and Cross-Lingual**. It uses **IndicNLP** to translate native queries into English SQL logic and operates entirely on local infrastructure (LM Studio/Ollama) to ensure total data privacy.
+
+## Features
+* **Feature 1: Hybrid Routing Engine** – Uses AutoGluon `TabularPredictor` to categorize queries before passing them to LLMs, ensuring high-speed, zero-hallucination routing.
+* **Feature 2: Supabase pgvector/PostGIS Integration** – A high-performance importer that enables both 768-dim semantic search and complex spatial queries (e.g., profiles within 200km).
+* **Feature 3: Multi-Turn Scientific Memory** – Implements `ConversationBufferWindowMemory` and DiskCache to maintain context throughout complex research sessions.
+
+## Tech Stack
+* **Frontend:** Streamlit
+* **Backend:** Python 3.10, LangChain, LM Studio, Ollama
+* **Database:** Supabase (PostgreSQL + pgvector + PostGIS)
+* **APIs / Services:** Local LLM API (localhost:1234), Local Ollama (localhost:11434)
+* **Tools / Libraries:** AutoGluon, IndicNLP, Pydantic, Pandas, NetCDF4
+
+## Project Setup Instructions
+
+```bash
+# 1. Install Dependencies
+pip install autogluon streamlit pandas requests pydantic langchain langchain-openai diskcache psycopg2-binary
+
+# 2. Setup Database Schema
+# Ensure 'new.env' is configured with your DIRECT_URL
+python argo_import_supabase.py --env-file new.env --data-dir "C:/path/to/data" --setup-only
+
+# 3. Import Data & Generate Embeddings
+# Requires Ollama running 'nomic-embed-text'
+python argo_import_supabase.py --env-file new.env --data-dir "C:/path/to/data" --embed
+
+# 4. Launch Application
+python -m streamlit run hyperpipeline/chat.py