divyam Prajapati

Architected and deployed an AI-powered Medical Scribe agent using LangGraph and AWS Bedrock, orchestrating multi-agent clinical workflows to autonomously generate structured Progress Notes from unstructured audio. Engineered a hybrid clinical coding pipeline integrating dynamic prompt engineering with medical domain knowledge bases, mapping raw clinical narratives to precise ICD-10 and CPT codes.

Designed and deployed a production-grade, stateful multi-step LangGraph agentic workflow to replace a legacy monolithic backend, achieving a 30% reduction in low-confidence evaluation metrics. Spearheaded development of an autonomous GenAI agent for intelligent routing and context-aware triaging, reducing human-in-the-loop escalation rates by over 50%.

Engineered an internal GenAI debugging assistant that executed automated stack-trace analysis and runtime context parsing, reducing Mean Time to Resolution for P2 bugs by 20%. Optimized a production GenAI recommendation engine using context-aware dynamic prompting and domain-specific fine-tuning, reducing model hallucination rates and suggestion errors by 16%.

Constructed an edge-optimized computer vision pipeline for real-time processing, implementing TensorFlow Lite quantization to reduce p99 inference latency by 75% on resource-constrained devices. Architected a high-velocity vector search layer using FAISS, achieving sub-50ms similarity search latency across an embedding database containing over 1 million records.

M.S. in Computer Science with a concentration in Intelligent Systems, completed in May 2025.

B.Tech in Computer Engineering with a concentration in Artificial Intelligence, completed in May 2023.