Avichai Ben

Avichai Ben is the Co-founder and CTO of Charm Security, an AI-powered Scam and Fraud Defense platform that protects organizations and their customers from scams, social engineering, and other human-centric fraud. Founded in 2024, Charm Security emerged from stealth with an $8 million Seed round led by Team8, positioning Ben as a key architect of next-generation fraud prevention technology.

Ben brings extensive expertise in AI-driven fraud detection solutions for Fortune 100 financial institutions. He holds a B.Sc. in Computer Science and Mathematics (summa cum laude, Hebrew University) and an M.Sc. in Computer Science (summa cum laude, Tel Aviv University), specializing in theoretical machine learning. His academic rigor is reflected in his research contributions: Ben has published research on machine learning at premier conferences including NeurIPS 2020 and AISTATS 2022, co-authoring papers on recurrent neural network optimization and efficient CNN learning.

Career history

• Co-Founder & CTO at Charm Security (Jan 2025–present)
• Entrepreneur in Residence (EIR) at Team8 (2024–2025)
• Head of Data Science at Transmit Security, where he led AI-driven fraud detection solutions for Fortune 100 financial institutions
• Senior role at Microsoft, driving security innovations for millions of users
• Senior role at Adallom (acquired by Microsoft for $320M), advancing cybersecurity technologies

What Avichai Ben could bring to his role as Co-founder and CTO at Charm Security: Ben's combination of academic rigor in theoretical machine learning, hands-on fraud detection leadership at Transmit Security serving Fortune 100 institutions, and experience at Microsoft-acquired companies positions him uniquely to architect AI-native fraud prevention systems. His deep understanding of behavioral patterns in human-centric fraud, coupled with expertise in building enterprise-grade detection systems, enables Charm Security to pioneer real-time intervention approaches that transcend traditional post-transaction fraud models.