At NeuroDiscoveryAI, I design AI-powered pipelines that turn unstructured EMR/EHR notes into analysis-ready datasets across neurological disease areas. I standardize clinical extraction schemas and develop evidence-based prompts informed by research literature and specialist input.
I built the ground-truth reference dataset for NeuroIntel, a retrieval-based clinical search system, and evaluated its responses against benchmark queries. I also contributed to NeuroAudit and AutoPrompter, validating outputs and supporting tool workflows across client projects.
Previously, at Jivi.AI, I designed evaluation frameworks for clinical AI modules and developed symptom-assessment logic. I also created a mental health feature and more than 50 workflows using an agentic AI platform.
