Karthik Sankaranarayanan

About Karthik Sankaranarayanan

Principal Machine Learning Scientist at Microsoft, where I lead AI innovation in Ads Trust & Safety—building intelligent systems that detect fraud, understand brands, and protect billions in advertising revenue• 14+ years bridging cutting-edge AI research with real-world impact: from publishing 2,400+ cited papers at top conferences to deploying production systems serving millions of users across Microsoft\'s advertising platform• I specialize in the full AI stack—foundation models, agentic systems, retrieval algorithms, and multimodal learning—with particular focus on adversarial robustness, fraud detection, and large-scale content moderation• Previously led IBM Research AI initiatives as Principal Investigator for the IBM-IIT Bombay AI Horizons Network, driving breakthroughs in multilingual NLP, low-resource languages, and enterprise conversational AI via IBM Watson• My work combines deep technical expertise (reinforcement learning, LLMs, computer vision) with strategic impact, helping organizations navigate AI adoption, build vs. buy decisions, and responsible AI implementation

Experience

1. Principal Scientist Manager

   Microsoft

   Sep 2021 — Present

   Lead AI and machine learning initiatives within Ads Trust & Safety, developing advanced AI agent systems for malicious advertising investigation, URL analysis, and real-time fraud detection across Microsoft\'s advertising platform• Architect domain-specialized foundation models and large language model (LLM) applications optimized for trust & safety, policy enforcement, and content moderation at scale, directly supporting multi-billion dollar fraud prevention efforts• Build retrieval systems using deep learning algorithms and relevance ranking models focused on brand understanding, semantic matching, and advertiser-query alignment to enhance ad quality and user experience across the advertising platform• Design comprehensive evaluation frameworks and benchmarking methodologies for agentic AI systems, leveraging reinforcement learning, test-time compute optimization, and multimodal learning to enhance detection precision• Drive cross-functional collaboration with product, engineering, and policy teams to deploy AI-powered solutions that protect advertisers and consumers while maintaining platform integrity and regulatory compliance

Education

• The Ohio State University

  Doctor of Philosophy - PhD, Artificial Intelligence

  2006 — 2011