Yamini Priya

I’m an AI Developer specialising in LLM applications, agentic systems, and RAG architecture. I focus on building AI that’s measurable, grounded, and production-ready—so performance improvements aren’t just claims, they’re verifiable.

I’ve built and deployed tools including a real-time multimodal interview agent, an adversarial RAG system, and a context engineering tool that achieves 90%+ noise reduction on codebases. My approach combines agentic design with evaluation-driven iteration, including multi-mode loops and relevance checks before generation.

In my AI Risk Research Extern role, I researched societal risks of AI—covering economic monopoly and power concentration as well as autonomous weapons and the AI arms race. I translated findings into case studies and mitigation recommendations, emphasizing balancing innovation with security and equitable access.

Through projects like Context Excavator, Vision Agent, and CrossLens, I’ve worked across real-time multimodal orchestration, AST-based codebase context mapping, and sceptical/adversarial retrieval. I’m energized by building systems where reliability comes from architecture, guardrails, and robust evaluation—not just prompting.