Saraswathi

• Developed LLM-powered enterprise AI applications using Azure OpenAI, LangChain, and Hugging Faceto automate document analysis and knowledge retrieval for banking operations.

• Designed RAG pipelines using vector embeddings and semantic search to retrieve knowledge from large financial document repositories.

• Built AI-driven REST APIs using Python and FastAPI to integrate machine learning and LLM models with enterprise banking platforms.

• Implemented high-performance vector search architectures using FAISS and Pinecone for scalable semantic retrieval.

• Designed agentic AI workflows using LangChain and LangGraph, enabling multi-step reasoning, document retrieval, summarization, and automated response generation for enterprise knowledge systems.

• Built Generative AI applications using Google Vertex AI and Gemini models, integrating embeddings, vector search, and LLM APIs for enterprise knowledge systems.

• Implemented automated ML model deployment and versioning workflows integrated with CI/CD pipelines to support reliable production AI systems.

• Designed multi-agent AI systems with Agent-to-Agent (A2A) communication, enabling collaboration between multiple AI agents for workflow automation.

• Built and optimized ML training, validation, and deployment pipelines using Azure Machine Learning and AWS SageMaker to support production AI workloads.

• Designed and implemented data ingestion and preprocessing pipelines using Python and SQL to prepare enterprise datasets for LLM-powered AI applications and model training workflows.

• Applied prompt engineering and context optimization techniques to improve LLM accuracy and reduce hallucinations in enterprise AI applications.

• Developed LLM evaluation pipelines using RAGAS and G-Eval frameworks to measure answer relevance, groundedness, and hallucination rates for Retrieval-Augmented Generation (RAG) systems.

• Implemented LLM fine-tuning and prompt engineering strategies for enterprise GenAI applications and integrated AI services with backend APIs and microservices.

• Deployed containerized AI applications using Docker and Kubernetes (AKS/EKS) enabling scalable and resilient AI model serving.

• Integrated Azure Cosmos DB with LLM applications to manage application state, conversation context, and structured data storage.

• Optimized embedding-based retrieval strategies to improve search accuracy and contextual relevance in LLM responses.

• Improved document retrieval accuracy by 35% by optimizing vector embedding strategies and semantic search techniques in enterprise RAG pipelines.

• Developed machine learning models for predictive analytics and enterprise decision systems using Python, Scikit-learn, and TensorFlow.

• Designed and implemented ETL pipelines using Python and SQL to process large enterprise datasets for machine learning training and analytics workflows.

• Built batch and streaming data pipelines to process structured and unstructured enterprise data for predictive analytics applications.

• Built data ingestion pipelines integrating relational databases, APIs, and external data sources support scalable ML data processing.

• Implemented feature engineering and data preprocessing workflows using Pandas and NumPy to improve model performance and training efficiency.

• Deployed containerized ML services using Docker and Kubernetes enabling scalable model serving in cloud environments.

• Developed RESTful APIs using Python and Flask to expose machine learning model predictions for enterprise systems.

• Implemented automated ML training and batch inference pipelines to support production machine learning workflows.

• Collaborated with data engineers, DevOps teams, and business stakeholders to translate business requirements into production-ready ML solutions.

• Implemented model monitoring and performance tracking frameworks to maintain reliability and accuracy of deployed ML models.

• Developed backend applications using Python and SQL to support enterprise business operations and reporting systems.

• Performed data extraction, cleansing, and transformation using Pandas and NumPy to prepare datasets for analytics and reporting.

• Assisted in building basic machine learning models (classification and regression) using Scikit-learn for internal analytics use cases.

• Worked on feature engineering and data preparation to improve data quality and support ML experimentation.

• Designed and developedRESTful APIs using Python and Flask to enable integration between backend systems and data services.

• Wrote optimized SQL queries and database procedures to efficiently handle data retrieval and storage operations.

• Collaborated with cross-functional teams (business analysts, QA, and developers) to gather requirements and deliver scalable solutions.

• Supported integration of analytics and ML outputs into applications to enable data-driven decision-making features.

• Performed unit testing, debugging, and performance tuning to ensure application reliability and efficiency.

• Used Git for version controland followed Agile and SDLC practices for development, testing, and deployment cycles.

• Participated in data validation and quality checks to ensure accuracy and consistency of data used in analytics and reporting systems.

• Assisted in performance optimization of applications and database queries, improving system efficiency and reducing data processing time.