Akash Kumar Agarwal

Title of the Talk: Emotion-Aware Customer Interaction: From Multi-Modal Research to Real Pharmacy Voice AI

Abstract :
Contact centers generate enormous volumes of affective signals like tone, pacing, hesitation, word choice, that conventional routing systems discard entirely. This talk presents EMCI-Net, a multi-modal emotion fusion architecture that combines speech emotion recognition (via a fine-tuned Wav2Vec 2.0 encoder) with BERT-based sentiment analysis through an attention-weighted late-fusion mechanism, achieving 89.3% weighted accuracy on the IEMOCAP benchmark, a 4.7-point improvement over the strongest single-modal baseline. Beyond accuracy, the architecture incorporates a Human-in-the-Loop retraining loop designed to keep model performance stable as call patterns and vocabulary drift over time.

The second half of the talk moves from research to practice. Drawing on direct experience implementing voice AI in a pharmacy contact center including outbound IVR for prescription pickups and refill reminders, an inbound conversational voice agent for refills, prescription status, and frictionless pharmacist transfer, and a quality agent that mines call transcripts for insights, the speaker maps each research component to its production counterpart, and is candid about where simulated research metrics stop and deployed, real-world experience begins.

Attendees will leave with a clear-eyed view of what multi-modal emotion AI can genuinely deliver in customer-facing systems today, and where the open research and deployment gaps still are.

Bio :
Akash Kumar Agarwal is a MarTech and Platform Architect with over 17 years of expertise in technology consulting, specialization in high-scale enterprise systems, and applied artificial intelligence research. In his current role at Albertsons Companies, Akash leads the design and implementation of advanced AI-driven voice and contact-center architectures, notably architecting production-grade voice AI solutions within regulated pharmacy and healthcare environments.
Alongside his industry practice, Akash conducts applied research on explainable AI (XAI) and emotion-aware AI systems, with findings published across IEEE conference venues. He remains active in the global academic community, frequently serving as a Technical Program Committee (TPC) chair, committee member, and peer reviewer for numerous IEEE and international engineering conferences.