Senior Data Scientist at Unilever R&D · Building production GenAI systems at scale

• Agentic AI & RAG — LangGraph multi-agent orchestration, retrieval-augmented generation, MCP integration
• Knowledge Graphs — Neo4j, graph-enhanced retrieval, ontology-driven pipelines
• LLM Evaluation — Langfuse observability, RAGAS metrics, production monitoring
• Document AI — Docling, OpenSearch, large-scale R&D document ingestion (~15K docs)

Pramāṇa — An open-source agentic RAG system built with Gemma 4, BGE-M3, LangGraph, Qdrant, Langfuse, and RAGAS.
Modernised from production architecture at Unilever R&D.

Python · LangGraph · LangChain · LlamaIndex · Neo4j · Qdrant · Azure AI
Gemma · BGE-M3 · Langfuse · RAGAS · Docling · FastAPI · Docker