Supriti Vijay is an AI Researcher at Cisco, where she pushes the frontier of efficient language models for AI security. A Carnegie Mellon graduate specializing in Language Technologies, her research asks a simple question: can specialized, efficiently trained language models outperform frontier-scale models on the tasks that matter? To answer it, she develops agentic post-training methods, builds compact models for security-critical applications, and designs evaluation frameworks that measure real capability, not just benchmark performance. Recognized through the AAAI Undergraduate Scholar and Adobe Women in Technology Scholar awards, her research sits at the intersection of AI security, trustworthy evaluation, and efficient AI systems. Outside of research, she enjoys swimming and is passionate about using technology for social good.