GPT Bot - Pilot for CPG Client & Phase II PROD for Kitchen and Bath Products ClientData Engineering:Executed analytical transformations and generated Views on Databricks Hive data, reducing storage requirements by eliminating data duplicationLeveraged Apache Spark & SQL to optimize data processing and improve overall efficiency.Generative AI:Developed text to sql bot with LLM model (GPT 3.5 Turbo) which will take the natural language as a user query and generate the SQL queryDeveloped Nudging prompt with NLP techniques which will Nudge the user if the query provided is incomplete which lead to increase in the accuracy of the bot by 7%Implemented the Question Rephrasing module with LLM model (GPT 3.5 Turbo) to take the previous 2 conversations to answer if there is any follow up questions in a conversationThis involved several prompts like context classifier, complexity classifier, data selector, text to sql generator, widget generator for charts etc. along with RAG and Context in Vector DB storeCreated multiple prompts/agents which additional context, few shot examples, RAG and several other techniques to fine tune and improve the accuracy of the moduleBackend Development:Developed multiple backend API endpoints with Fast API framework leveraging various HTTP methods including GET, POST, PUT, and DELETEImplemented JWT authorization and user ID validation from the database to enhance securityDeveloped CI/CD pipelines in Azure DevOps repo for integration and deployment along with docker image buildImplemented user logs view feature for admins and super users

Global Health Dashboard Micro Service ArchitectureData Engineering:Built data pipelines for ETL processes with Azure Data Factory, including setting up and managing timely schedule triggersimplemented Email notifications for Data pipelines with Azure logic appsUtilized Azure Data Lake Storage (ADLS) and Databricks for processing and transforming data; stored final analytical data in MS SQL Server databases.Calculated additional KPIs and metrics based on YTD, Period, MAT and YearBackend development:Created multiple backend API endpoints with the Azure function apps, leveraging various HTTP methods including GET, POST, PUT, and DELETEImplemented JWT authorization and user ID validation from the database to enhance securityImplemented microservice Architecture for accessing several applications at one central repositoryImplemented RBAC (role based access control) Leve 1, Level2 and Level 3Developed CI/CD pipelines in Github repo for integration and deployment

Consumer Analytics (CA) Dashboard | Idea Engine Web ApplicationBackend Development:Created multiple backend API endpoints using the Flask framework, leveraging various HTTP methods including GET, POST, PUT, and DELETEAdministered JWT authorization and user ID validation from the database to enhance securityDeveloped CI/CD pipelines in Azure devOps repo for integration and deployment using docker to build imageData Engineering:Built data pipelines for ETL processes using Azure Data Factory, including setting up and managing timely schedule triggers.Utilized Azure Data Lake Storage (ADLS) and Databricks for processing and transforming data; stored final analytical data in MS SQL Server databases.Performed web scraping with Beautiful Soup, a Python library, to extract data from the internet, successfully scraping over 80,000 articles and storing them in a database for insight generation in the chatbotPerformance OptimizationOptimized the Azure web app performance by implementing Auto scaling of instances thereby reducing the cost by 18%