Partha Kuila

Project name: MCP Conversational Voice/Chat in Nationwide Building Society

• Architected and deployed highly scalable Amazon Connect contact canter solutions, configuring complex IVR work flows, queues, routing profiles, and dynamic routing strategies for enterprise clients
• Optimized Amazon Lex conversational AI by designing sophisticated intents, slots, and dialog flows with intent disambiguation, significantly improving IVR automation rates and Natural Language Processing (NLP) accuracy.
• Integrated Amazon Bedrock foundation models to provide AI-powered agent assistance within Amazon Connect
• Implemented Retrieval-Augmented Generation (RAG) using Bedrock and enterprise knowledge bases for accurate customer responses.
• Configured Bedrock Guardrails to filter harmful or sensitive content before presenting responses to agents.
• Integrated AI-powered Call Summarization and Amazon Connect Screen Recording, boosting agent pro ductivity and streamlining quality assurance processes.
• Automated call data analysis by storing CTR logs in Amazon S3 and querying them using Amazon Athena.
• Streamlined operational reporting by utilizing Amazon Kinesis Data Streams to ingest and route outbound cam paign events to AWS Glue for processing and cataloging.
• Implemented AWS analytics with Athena, S3, DynamoDB, Lex, CloudWatch, and QuickSight to strengthen reporting and automation processes.
• Reduced deployment risk by running ML pipelines through MLflow and SageMaker CI/CD with monitoring.
• Accelerated conversation understanding by summarising agent-customer exchanges using Bedrock GenAI with Titan G1 Express.
• Analysed customer complaints and trend patterns to enhance insights for conversational banking solutions.
• Elevated customer service operations by delivering end-to-end Amazon Connect solutions.
• Fixed configuration issues in Amazon Connect and contact flows to improve reliability of contact centre operations.

Project name: North Star Intelligent Talent Optimization for Workforce Management

• Supported profitability goals by enabling Northstar capacity management and supply-gap optimisation for bench reduction.
• North Star is innovative in its end-to-end approach to making optimal recommendations for the business to hire, procure contractors or reskilling, starting with demand and supply forecasts based on various internal and external data sources and sensors
• Contribution: Demand & Supply LSTM model, Bayesian modelling, Ensembling through voting mechanism, Markov Chain Model, T2G Alignment, JRS Forecasting model, Band Pyramid, Scenario Analysis, Recommendations.

• Built final Auto-ML model to generate dependable predictions.
• Used Python with NLTK, spaCy, and pandas for text processing.
• Applied R and H2O to run the full modelling workflow.
• Analysed 550 data points through feature selection to identify 50 key features.
• Imputed missing values and removed constant and quasi-constant features.
• Classified data inputs to enhance performance metrics across personal and business data.
• Extracted keywords from employee feedback to create performance scores for analysis.

• Developed ARIMA model for forecasting optical line terminal capacity, supporting strategic planning decisions.
• Automated weekly execution of daily OLT capacity predictions, enhancing operational efficiency.
• Analysed customer complaint data to identify patterns, guiding targeted project enhancements.

• Analysed complex data sets to identify trends, patterns, and insights for informed decision-making.
• Monitored data collection results to identify trends and anomalies, ensuring data integrity for ongoing projects.
• Analysed village and city areas through segmentation of census datasets to inform urban development strategies.
• Processed satellite imagery data for analysis, contributing to informed planning and development decisions.
• Presented findings and recommendations to senior management through clear, concise reports and presentations.