Reshwanth Mandala

Architected and operationalized RAG-based agent systems using Vertex AI, Gemini models, BigQuery, LlamaIndex, FAISS, and Pinecone, achieving 99% retrieval accuracy across large enterprise knowledge bases. • Built and deployed production-grade AI agents with human-in-the-loop (HITL) controls, audit logs, traceability, and fallback mechanisms to meet governance, risk, and regulatory compliance requirements. • Developed agent-backed microservices using Python, C#, FastAPI, and REST APIs, enabling secure integration with CRM, ERP, procurement, and telemetry systems for autonomous and tool-driven agent execution.

Implemented automated ML pipelines (CI/CD) with Jenkins and GitLab, reducing manual deployment errors by 30% and accelerating model updates for production systems. • Engineered high-performance distributed data processing pipelines using Spark MLlib, PySpark, Dask, Hadoop, Hive, and BigQuery, enabling real-time processing of millions of records with low latency. • Developed real-time predictive analytics for labor forecasting, dynamically integrating streaming data from market trends and customer behavior, improving forecast accuracy by 15%.

Built and optimized deep learning models with Keras, TensorFlow, and PyTorch to predict claim severity and policyholder risk, improving underwriting accuracy and loss ratio forecasting. • Harnessed big data technologies such as Hadoop and Spark to process large-scale policyholder and claims data, enabling accurate premium forecasting and efficient resource allocation. • Managed SQL and NoSQL databases including MySQL, MongoDB, and PostgreSQL to extract operational insights that optimized claims adjudication workflows and provider network management. • Implemented anomaly detection systems for fraud detection, using statistical models to flag suspicious claims patterns and reduce fraudulent payouts.