sage-climate-agent-orchestrator

Multi-agent climate data pipeline with MCP servers for SAGE Lake access, n8n workflow automation, and LLM-powered climate-health report generation

SAGE Data Lake (1.78B rows)     Processing Layer              Application Layer
+-------------------------+    +----------------------+    +-------------------+
| ERA5 Climate (362M)     |    | Feature Engineering  |    | REST API          |
| WHO Health (290M)       |--->| Model Training/Infer |--->| Real-time Stream  |
| IHME GBD (47M)          |    | Knowledge Graph      |    | Dashboard UI      |
| 268M Vector Embeddings  |    | Agent Orchestration  |    | Alert System      |
| 33M Causal KG Triples   |    | Evaluation Pipeline  |    | Report Generator  |
+-------------------------+    +----------------------+    +-------------------+

| Source | Records | Domain | Usage | |--------|---------|--------|-------| | WHO GHO | 190M+ | Health | Disease indicators across 194 countries | | ERA5 CCD | 362M+ | Climate | Daily climate reanalysis (temperature, precipitation) | | IHME GBD | 47M+ | Health | Global burden of disease estimates | | World Bank WDI | 24M+ | Economics | Socioeconomic covariates | | UNICEF | 8.8M+ | Child Health | Immunization, nutrition, WASH indicators | | OpenDengue | 5.7M+ | Epidemiology | Dengue case counts across 129 countries | | OECD Pharma | 19.6M+ | Pharmaceutical | Drug consumption, pricing, expenditure | | CARD AMR | 271K | Genomics | Antimicrobial resistance genomes | | WorldPop COGs | 200K+ | Geospatial | Population density rasters at 100m/1km |

# Backend
pip install -r requirements.txt
python -m src.main

# Frontend (if applicable)
cd frontend && npm in