Jaya Sri

I designed and delivered enterprise-grade GenAI solutions focused on scalable document intelligence and scientific workflows. I led the development of FAISS-powered QA bots and retrieval-augmented generation (RAG) pipelines, significantly reducing manual document review while improving semantic search accuracy across large internal datasets. I fine-tuned large language models using LoRA and QLoRA to optimize GPU utilization and reduce training time, deploying low-latency inference services using AWS SageMaker, Lambda, and API Gateway. I built real-time monitoring APIs with FastAPI and PostgreSQL to track token usage and system performance, improving cost visibility and operational reliability. I also implemented HIPAA-compliant fine-tuning and deployment workflows using AWS KMS, IAM, and encrypted S3, standardized inference schemas with Pydantic, and built embedding and feature stores using DynamoDB and Redis to accelerate production RAG systems. I supported multi-modal GenAI projects and mentored engineers on LLM lifecycle and MLOps best practices.

I built and deployed Generative AI copilots that enhanced financial research workflows by enabling intelligent document summarization and faster insight discovery for analysts. I developed LangChain-based prompt pipelines and context enrichment workflows to generate accurate, domain-specific investment narratives. I trained and optimized financial-domain LLMs using distributed training, LoRA, and quantization techniques, improving inference efficiency while maintaining high factual accuracy. I implemented retrieval-augmented generation (RAG) pipelines using Pinecone vector search, significantly improving semantic retrieval across proprietary financial datasets. I ensured model compliance and reliability by conducting audits with SHAP and LIME, implementing PII filtering, and integrating scalable inference infrastructure on AWS ECS with monitoring using CloudWatch and Prometheus. I also automated MLOps workflows using GitHub Actions and MLFlow and built executive dashboards to track model performance and system health.

I worked on high-scale retail systems, building real-time inventory data pipelines using Azure Data Factory and Azure SQL to improve product availability and reduce stockouts. I developed and deployed .NET Core microservices for customer tracking and returns, supporting scalable and cost-efficient backend services. I improved application performance by integrating Redis caching and optimizing Elasticsearch indexing, resulting in faster page loads and improved search relevance. I modernized legacy services by containerizing applications and deploying them to Azure Kubernetes Service (AKS), improving scalability and operational resilience. I also built CI/CD pipelines using Azure DevOps, delivered responsive React.js components integrated with backend APIs, and implemented automated testing using Selenium and Playwright. I contributed to event-driven checkout architectures and supported business reporting with Power BI dashboards.