Smit-Parekh/proactive-supply-chain-disruption-agent

**Technical Summary:** The agent orchestrates event ingestion from real-time data streams (GDELT, weather APIs) through a stateful LangGraph workflow that performs client-specific risk assessment via PEFT-fine-tuned Gemma-3-1B, hybrid impact forecasting combining XGBoost delay predictions with LLM qualitative reasoning, and RAG-enhanced mitigation synthesis from vector-indexed SOPs and historical examples. End-to-end MLOps integration with Vertex AI Pipelines, custom training jobs, and MLflow experiment tracking enables reproducible model iteration and production monitoring across client cohorts (pharma/petroleum archetypes with distinct sensitivities).

No commits in the last 6 months.