Amar Viswanathan

I’m a Senior Applied Scientist at Amazon, where I lead the design and deployment of large-scale ML and agentic AI systems in Amazon Ads — systems that operate across tens of millions of ad campaigns worldwide. My current work focuses on autonomous optimization: building systems that reason, adapt, and act at scale with minimal human intervention.

Before Amazon, I was Principal Scientist at SAP, where I led the end-to-end research and launch of SAP Consultant Capability (SCC), a production enterprise RAG system handling 22,000+ multi-turn conversations per day with measurable business impact. I’ve also worked at Dataminr, building real-time LLM systems over 19 million tweets per day, and at Siemens Research, where I served as Co-PI on a $1.8M DARPA program building multimodal knowledge graphs.

My work sits at the intersection of agentic AI, RAG, knowledge graphs, and production ML — turning research ideas into systems that ship and scale. I hold a PhD from Rensselaer Polytechnic Institute under James Hendler, have published at ECIR, CIKM, and AAAI, and hold a granted US patent with several more in progress. I write and share about what I’m learning at the frontier: agentic architectures, GraphRAG, and reasoning in LLMs, and the messy gap between research and real-world deployment.