Sumanth Vaidya

• As an AI LLM Technology lead at Accenture, my responsibilities include, identifying new Machine learning/Large Language Model automation opportunities, perform tech assessment, propose best comprehensive solution with end-to-end solution design & architecture, business case and leading development through deployment for large business transformation deals under Accenture Operations
• In span of last 8 years, I successfully designed and delivered automation solution for over 30 deals in Accenture operations using AI/ML/LLM based approaches across multiple domains include – Finance & Accounting, HR, Procurement, IT Helpdesk, Health Care yielded efficiency of over 250 FTE savings for Accenture.

LLM based platform for Pharmacovigilance

• Designed comprehensive product suite for pharmacovigilance automation using LLM and enterprise standard integrations with Oracle Safety systems, medical coding applications and upstream systems
• Designed and crafted Large Language Models (LLMs) solution approach and prompts using Azure Open AI -GPT 4, comprehending complex medical narratives
• Delivered scalable and reusable platform for Pharmacovigilance using Azure cloud services and container-based architecture using Kubernetes
• Successfully guided project from proof of concept (POC) through full production deployment, ensuring seamless implementation for largest European Pharma company, resulting in 15 FTE savings
• Potential 8 clients are in pipeline which could yield up to 120 FTE savings.

Insurance benefits comparison

• Designed and developed LLM based solution for comparing insurance provisions & benefits and deriving best benefit for customer as per legislative guidelines using Azure Open AI GPT4
• Solution is successfully integrated and deployed for one of largest Insurance providers in North America
• Automation has delivered efficiency of over 60% with 8 FTE savings for Accenture Operations.

Dispute Management

• Led automation for dispute management process with multilingual support for largest medical device company using GPT 3.5 and Azure AI services.

Procurement and HR

• Demonstrated Proof of Concept (POC) for FAQ based automation using Retrieval Augmented Reality (RAG), Embeddings, Vector Search for HR helpdesk.
• Automated order management process integrated with RPA bots using Large Language Models for one of world's leading food and beverages company.

Unified Helpdesk/Email Advisor Platform

• Designed and delivered end to end comprehensive email advisor solution for touchless helpdesk using Machine Learning, NLP and Large Language models, integrated with enterprise ticketing platforms like ServiceNow, and enterprise ERP's -SAP/Oracle via RPA's and REST APIs
• Designed a reusable application framework which can orchestrate Machine learning model/LLM prompts as service for email classification, extraction, vendor authentication and integration with ERPs for automated resolution
• Operationalized as a scalable platform with container-based architecture using Azure Kubernetes deployed over 20 clients yielded benefit of up to 120 FTEs in last 5 years
• Designed flexible integration provision for ERP's/ticketing systems for RPA and REST APIs based on client convenience for seamless integration
• Ideated and implemented advance concepts- Agent assist, Customer 360 to improve overall helpdesk efficiency and customer satisfaction.

Helpdesk Quality Audit platform

• Developed and deployed Deep Learning models using Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) architectures to analyze customer sentiment and emotion from client emails
• Integrated results into Power BI Dashboards, streamlining service desk quality auditing and saving significant man-hours.

• As a senior developer I worked on support and maintenance of Mimictron tool, which crawl websites and download financial reports and image processing/OCR (Optical Character Recognition) to identify the dynamic UI elements.

• Intelligent Text Analysis Platform for Lease Abstraction – Developed end to end workflow for automated lease document processing using machine learning approach.
• Implemented machine learning solution for extracting key lease information form lease documents using Python and scikit learn and exposed REST APIs for integration with multiple clients.

• Collateral Management and validation- Worked as Java developer in developing an integration framework for collaborating with upstream and downstream systems for payment data exchange using Java
• Integrated GP2 distribution system through Webservices.