* Implemented core platform logic using Java and RESTful APIs to power the MarketFusion quotation flow, ensuring reliable data exchange and smoother user interactions

* Delivered full-stack features using Java and React, enhancing page responsiveness, form performance, and end-to-end usability in close collaboration with product and leadership

● Led a team of 5 in collaboration with Data Scientists to develop an automated dashboard integrating APIs for data retrieval and processing from AWS S3, enabling clients to visualize data insights and enhancing customer engagement by 20%

● Architected a customer-facing chatbot powered by OpenAI GPT-4 and LangChain with Pinecone vector DB, utilizing Generative AI and LLMs to improve conversational accuracy and scalability, now serving over 25+ clients

● Utilized LangChain with Pinecone vector database to enable semantic search and enhance chatbot accuracy; improved real-time query resolution and achieved a 30% boost in user engagement

● Conducted A/B testing on chatbot response strategies, comparing prompt-only and retrieval-augmented methods; optimized answer quality and boosted user engagement across client deployments

● Developed a dynamic employee management dashboard using React and TypeScript, enabling managers to track progress and assignments for 30+ projects across 8 teams, supported by a scalable Node.js backend

● Designed and implemented APIs with Django to facilitate optimized data retrieval for project and employee management systems

● Optimized database schemas and queries, reducing response times by 25% and ensuring smooth backend integration

● Implemented OAuth 2.0 for API security; added retry logic, timeout handling, & token validation to enhance microservices resilience

● Crafted an intuitive user interface using HTML, CSS, and React to develop a web page facilitating policy initialization, and seamless interaction using Redux and NodeJS

● Engineered Node.js backend code to validate policy initialization forms and PDFs through integration with PDF verification API

● Deployed a queuing mechanism within Redis to handle customer requests, elevating the authorization process and ensuring streamlined management of incoming data

● Employed ReactJS PDF packages and crafted Python scripts to automate default account creation. Utilized advanced socket communication and Redis to facilitate smooth video communication between clients and customer service

● Implemented Docker for cross-platform compatibility, employed Kubernetes for efficient container management and scaling, and seamlessly integrated GitHub with Jenkins to establish a robust CI/CD pipeline

● Configured MongoDB database for storing policy initialization data, ensuring secure and efficient data retrieval. Employed Mongoose for Node.js to interact with the database and enforce schema validations

● Revamped transformer-based models including BART, T5, and PEGASUS on COVID-19 articles and research papers to generate human-like summaries of ~2TB of data

● Optimized data preprocessing using advanced NLP methods such as tokenization and lemmatization, resulting in a refined dataset

● Implemented TF-IDF vectorization to transform raw text into quantitative features, boosting ML model accuracy by 12%

● Enhanced NLP capabilities by fine-tuning PEGASUS model leveraging the NVIDIA DGX workstation

● Outperformed the initial PEGASUS model demonstrating superior accuracy and Rouge scores by 35%

● Employed CI/CD for automated testing and deployment of fine-tuned transformer models, ensuring rapid and reliable model