Hire Sravan G. - Generative AI Engineer

Designed LLM-driven risk summarization pipelines using GPT-3.5 and AWS Lambda, reducing underwriting turnaround time by 50% and significantly improving document processing efficiency through cross-functional collaboration. Built secure insurance claims summarization systems on Amazon SageMaker using attention-based models trained on proprietary datasets, capturing domain-specific terminology and risk signals. Developed OCR-based document ingestion pipelines with AWS Textract, integrating GenAI-powered Q&A systems to automate claims triage and large-scale document analysis. Automated customer support email workflows using LangChain agents integrated with CRM platforms, reducing response times while maintaining regulatory compliance and contextual accuracy. Created structured prompt orchestration pipelines using OpenAI models and fine-tuned FLAN-T5 to generate clear, compliant explanations of policy terms, improving customer transparency and call-center resolution rates. Engineered retrieval-augmented conversational agents (RAG) using ChromaDB and Cohere to surface insights from historical claims data and underwriting guidelines, enhancing decision accuracy and user experience. Implemented reinforcement learning with human feedback (RLHF) pipelines driven by claims agent reviews, continuously improving response quality and alignment with business expectations. Integrated AWS Comprehend for sentiment analysis and named-entity recognition on customer feedback, enabling adaptive engagement strategies and more responsive GenAI interactions. Designed secure, human-in-the-loop GenAI workflows for claims approvals, ensuring auditability, bias mitigation, and regulatory compliance in high-impact automation scenarios. Led prompt optimization workshops with underwriting SMEs, applied knowledge distillation to reduce inference latency, and monitored model quality using ROUGE and BERTScore metrics.

I partner closely with product managers, data engineers, and business stakeholders to translate complex requirements into production-ready AI/ML solutions, delivering predictive analytics, recommendation systems, and document intelligence capabilities through structured, end-to-end ML delivery frameworks. My work spans time-series modeling, computer vision, NLP, and recommender systems, with a strong emphasis on scalability, interpretability, and measurable business impact. Key contributions include: Designed and deployed time-series forecasting and anomaly detection pipelines using LSTM-based sequence-to-sequence autoencoders (TensorFlow/Keras) to predict Remaining Useful Life (RUL), reducing unplanned maintenance events by 18% Built computer vision solutions using CNNs and Mask R-CNN for object detection and image classification, supporting automated insurance claims processing and vehicle damage assessment Applied autoencoders and variational autoencoders (VAEs) for image reconstruction and denoising, improving the quality and usability of scanned claims and policy documents.

* Designed and developed scalable backend and GenAI-driven systems for audit, compliance, and insurance platforms, focusing on performance, security, and intelligent automation. * Built and maintained RESTful APIs using Spring Boot and Jersey to support audit and compliance workflows with high availability and low latency. * Architected backend services for real-time audit data ingestion, processing, and reporting, enabling faster insights and regulatory compliance. * Integrated MongoDB for scalable, document-based storage and optimized data access for large-volume transactional workloads. * Implemented caching strategies and performance optimizations to support high-traffic systems and reduce response times. * Developed secure authentication and authorization mechanisms, ensuring data protection and compliance with industry standards. * Conducted code reviews and refactored services to improve scalability, maintainability, and system performance. * Built OCR-based document ingestion pipelines using AWS Textract, integrating them with GenAI-powered Q&A systems to automate claims triage and large-scale document analysis. * Automated customer support email workflows using LangChain agents integrated with CRM platforms, significantly reducing response times while maintaining compliance and contextual accuracy.