Vennela C.

Designed and developed Agentic AI systems for banking operations using multi-agent architectures to automate complex workflows such as fraud investigation, customer issue resolution, loan eligibility assessment, transaction dispute handling, risk analysis, and personalized product recommendation generation. • Built multi-agent pipelines using LangChain, LangGraph, and CrewAI to orchestrate LLM interactions, banking business rules, core banking APIs, CRM systems, fraud detection platforms, and transaction monitoring tools. • Developed enterprise-grade RAG pipelines integrating vector databases and document repositories to enable context-aware question answering over banking knowledge bases, including lending policies, KYC/AML regulations, credit card rules, account procedures, transaction policies, and customer support documentation. • Implemented vector search and embedding pipelines using FAISS, Pinecone, and ChromaDB for semantic retrieval of banking-specific content such as loan eligibility rules, product catalogs, customer interaction history, fraud scenarios, and operational procedures. • Designed end-to-end GenAI pipelines for banking use cases including ingestion of customer call transcripts, transaction logs, fraud alerts, support tickets, and financial product catalogs, followed by chunking, embedding generation, retrieval, and response synthesis. • Integrated foundation models through Amazon Bedrock using Claude and Titan models to support banking use cases such as intelligent customer support, fraud incident summarization, personalized financial recommendations, compliance monitoring, and conversational self-service. • Developed conversational AI systems and banking virtual assistants integrated with enterprise APIs and backend systems to automate customer inquiries related to account balances, loan applications, credit cards, transaction disputes, fraud alerts, payment status, branch appointments, and product recommendations.

Developed machine learning models using Scikit-learn, TensorFlow, and PyTorch for banking use cases such as credit risk scoring, fraud detection, loan default prediction, customer segmentation, and transaction classification. • Built early-stage LLM-based applications using API integrations for banking workflows including automated customer support, financial document summarization, regulatory Q&A, policy interpretation, and knowledge management. • Designed custom retrieval-based solutions to improve search and access across banking documents such as loan agreements, KYC guidelines, compliance policies, transaction procedures, and customer service knowledge bases. • Implemented reinforcement learning models using OpenAI Gym for banking optimization problems such as portfolio allocation, fraud investigation prioritization, dynamic credit limit adjustment, and collections strategy optimization. • Developed large-scale data pipelines using Apache Spark and SQL-based processing systems to ingest and transform transaction data, customer profiles, account activity, and financial records from multiple banking platforms.

Developed transformer-based NLP models using BERT and GPT architectures for document classification, summarization, and semantic search. • Built RAG-based retrieval systems integrating structured databases and unstructured documents using LangChain and LlamaIndex. • Designed ingestion pipelines to process data from PDFs, APIs, and cloud storage into vector representations. • Implemented prompt engineering techniques to improve LLM outputs for enterprise use cases. • Built scalable ML pipelines using Azure ML, Kubeflow, and MLflow for model lifecycle management. • Developed distributed data processing pipelines using Apache Spark, Spark SQL, and Hadoop ecosystem tools.

• Designed and developed scalable batch and real-time data pipelines using Apache Spark, PySpark, and Apache Airflow. • Built real-time streaming systems using Apache Kafka and Spark Streaming. • Developed ETL pipelines using Python, Scala, and Informatica for data transformation and integration. • Designed data lake architectures using AWS S3 and implemented warehousing solutions using Redshift and Athena. • Orchestrated workflows using Apache Airflow DAGs for scheduling and monitoring.