AI Intern, Darwix AI (Multi-source RAG, voice-to-insight, agentic workflow automation)
Created agentic automation utilities that transform live voice transcriptions into structured downstream actions for enterprise workflows. Integrated a voice-to-insight pipeline that performs real-time transcription followed by chunked summarization and intent extraction. Used grounded LLM orchestration across multi-source retrieval to improve accuracy while reducing hallucinations during conversational intelligence generation.• Parsed live voice transcriptions into structured action outputs.• Performed transcription-driven summarization and intent extraction for CRM automation.• Reduced manual processing effort by approximately 40% for enterprise clients.• Improved end-to-end response accuracy and reduced query latency by optimizing retrieval and reranking.