Responsibilities include:

• Introducing GuestIQ to Four Seasons: Led the development and introduction of GuestIQ, an AI-powered IP that integrates multiple models to predict guest preferences, enhancing customer experience and driving revenue.
• Introducing Spotlight: AI Agent for Guest Intelligence & Next-Best Action: Designed and deployed a production-grade agentic LLM system using LangChain for orchestration, advanced prompt engineering, and retrieval-augmented generation (RAG) with embeddings and vector search to ground outputs in enterprise data. Implemented multi-step reasoning, tool use, and conversational memory to autonomously synthesize signals from diverse sources into real-time guest briefings and personalized next-best actions. Developed structured prompts, evaluation pipelines, and guardrails to ensure reliable, context-aware, and privacy-aware outputs. Significantly reduced manual profile review time while improving service precision, operational efficiency, and guest satisfaction.
• Introducing FS Text Tools — Unified AI Text Processing API (MCP-enabled): Designed and deployed a centralized, MLflow-powered API endpoint to serve as a Model Context Protocol (MCP) tool for AI agents, consolidating advanced unstructured data processing capabilities—including translation, PII and health-data redaction, language detection, sentiment analysis, and content safety—into a single scalable service. Enabled consistent governance, privacy-aware handling of sensitive data, cost optimization, monitoring dashboards, automated alerts, and simplified integration across applications and agent workflows.
• Guest Preference Modeling with NER: Built a DeBERTa-xlarge NER model to extract guest preferences from unstructured multi-channel customer touchpoints (e.g., chat, phone calls), enabling hyper-personalized experiences and tailored recommendations for future visits.
• Developing Recommendation Engine: Building a recommendation system using collaborative filtering, content-based models, and hybrid techniques to personalize property, room, and ancillary recommendations. Leveraging Databricks, Spark, Azure Synapse, and Azure SQL for large-scale processing and real-time inference.
• Sentiment Analysis Model: Developed and fine-tuned a BERT-based sentiment analysis model using Hugging Face and PyTorch to classify guest sentiment across multi-channel interactions (chat, email, phone transcripts). Achieved over 90% accuracy on labeled data; model insights were used to detect service gaps, monitor guest satisfaction trends, and inform personalized engagement strategies.
• MLOps & Model Deployment: Automating model training, versioning, and deployment using Azure ML, MLflow, Databricks, and Azure DevOps with CI/CD pipelines for seamless production integration.

Responsibilities include:

• Partnered with business and technology teams to identify business problems, built insightful analytical solutions and developed state-of-the-art solutions aligned with business objectives. Provided data warehouse & BI technology vision, roadmaps, blueprints, and project plans
• Extract, Transform and Load data from various sources to Azure Data Lake Storage using a combination of Azure Data Factory, T-SQL, Spark SQL, and U-SQL Azure Data Lake Analytics. Data Ingestion to one or more Azure Services - (Azure Data Lake, Azure Storage, Azure SQL, etc.) and processing the data in Azure Databricks
• Developed and implemented API services and monitoring systems in an agile environment for connecting and querying AAS (DAX, MDX) using .Net Core, ADOMD.NET / Entity Framework, ASP.NET MVC, ASP.NET WEB API, JWT, Microsoft SQL Server, Azure Blob Storage, Data Pipeline, Azure Logic App, and real-time Power BI
• Designed and executed well-engineered, easy-to-maintain, and reliable applications, using AI models, and helped to free up over 20 labor hours per week and facilitated users’ access provisioning requests
• Developed internal analytics and data exploration services, using ServiceNow Rest API, to analyze patterns and reduce ServiceNow ticket volume by flagging similarity across tickets using cluster analysis, automating ticket assignment, and predicting time to resolution for each request using regression analysis
• Utilized MS SQL, Azure Analysis Services, data warehousing programs, Power BI, and PowerApps for data visualization and analysis; Designed, created, and maintained automated visualizations for internal and client-facing reporting deliverables
• Owned and managed all changes to the data models and used Git Repository as a version control tool. Created data models, solution designs, and data architecture documentation for complex information systems

Responsibilities include:

• Provided executives with analytics and decision-support tools to help with strategic initiatives, roadmaps & process models
• Leveraged platforms such as Google and Adobe Analytics, besides MS SQL, Tableau, and other visualization toolsets for data intelligence and analysis, to gain data-driven insights into customer behavior and used for business development strategies
• Implemented Total Business Excellence [TBE™] model to assist businesses to achieve a sustainable competitive advantage by developing a culture that is integrated, aligned, and responsive to market and customer demands

Responsibilities include:

• Project 1: Conversational Agents

• Designed and developed AI-enabled agents using JavaScript, HTML, and API calls and used these as virtual advisors in financial services to provide a personalized experience across different digital channels and customer touchpoints
• Collected, studied, and interpreted large datasets using surveys and interviews; performed extensive data analysis to unveil User-AI interactions; investigating the adoption of these technologies by organizations and their acceptance by customers
• Conducted advanced statistical analysis using Python and R (e.g., structural equation modeling, multiple/logistic regression, mediation/moderation models, ANCOVA, MANCOVA, and linear mixed models) to test hypotheses; tested for the validity and reliability of constructs using lavaan, semTools, psych, dplyr, ggplot2, car, and emmeans libraries

• Project 2: Intelligent E-mail Management System (IEMS)

• Created a personalized AI platform and implemented supervised clustering algorithms such as Naïve Bayesian classification, Support Vector Machine, K-Nearest Neighbor, and Neural Networks to group received emails based on their importance and urgency
• Evaluated model performance and monitored the training process, eventually obtained excellent accuracy of 93%

• Teaching

• Instructed Business Intelligence, Business Process Model and Notation, App Development, ERPsim
• Used Multiple-Criteria Decision-Making (MCDM) models (e.g., analytic hierarchy processing (AHP), ELECTRE, VIKOR, TOPSIS, and CORPAS) as computational tools in evaluating multiple, contradicting, qualitative and quantitative criteria