Nadigadda Shiva Sai

Built a RAG pipeline over company internal documents to solve the context selection problem, chunking documents into semantically coherent sections, embedding them with dense and sparse vectors, and retrieving only query-relevant chunks via hybrid search before injection into the LLM prompt. Engineered production-grade LLM systems with context management strategies (conversation history pruning, sliding window summarization, dynamic prompt assembly) and agentic tool-calling pipelines featuring autonomous feedback, retry loops, and self-correction, enabling reliable AI agent execution under token constraints. Built a session-based AI document intelligence platform in Python and Google ADK that ingests PDF, DOCX, CSV, XLSX, and zip uploads, normalizes unstructured content into Markdown, and enables grounded question answering with citations through a retrieval specialist agent. Built a real-time network topology and observability dashboard using Next.js (React) to visualize connectivity across thousands of RTSP camera streams (Cameras NVRs Bridgedevice Cloud), enabling instant fault detection and reducing debugging time. Built scalable backend services and REST APIs using FastAPI, designing relational data models and monitoring pipelines for device health tracking, while implementing network integrations, firewall configurations, and camera onboarding to Bridge devices for reliable distributed system performance.

Built and deployed a multi-turn RAG pipeline using LlamaIndex and OpenAI GPT, enabling conversational querying over scientific PDFs with chunk-level source attribution, the model cites the exact document and section behind every response. Designed scalable REST APIs using FastAPI, Node.js, and Python, supporting high-volume CRUD operations and data-intensive backend services. Designed and deployed APSI a web surveillance platform using React.js and next.js, improving system reliability by 40% through optimized backend architecture and secure data pipelines. Provided hands-on mentorship to a team of 45 junior developers through structured code reviews and proactive blocker resolution, while defining project scope, deliverables, timelines, and resource requirements to align client expectations with engineering capabilities. Built ETL pipelines using Python to automate data ingestion from Excel into MySQL, reducing manual effort by 70% while ensuring data integrity and audit compliance. Implemented Role-Based Access Control (RBAC) and JWT-based authentication, improving platform security by 60%. Integrated unit, integration, and stress testing into Jenkins CI/CD pipelines, enabling reliable, scalable deployment of AI and backend systems for multi-user environments.

Developed RESTful and SOAP web services with OpenAPI specifications, ensuring robust and well-documented backend interfaces. Built responsive UIs using React.js, TypeScript, and Material UI, ensuring cross-browser compatibility and seamless user experience. Achieved 90%+ test coverage using JUnit, Mockito, MockMvc, and Karma/Jest reducing regression issues by 30%. Automated deployment pipelines with Jenkins, Docker, and AWS using container orchestration and Kubernetes pod monitoring. Performed static code analysis with SonarQube and security assessments with Veracode, meeting enterprise security standards.

Migrated web application from ReactJS to Next.js, improving page load speed by 45% and significantly boosting SEO performance. Built a role-based Management System for administrative workflows, improving data security by 40% and operational efficiency by 25%. Developed a Java API layer for dynamic MongoDB schema extraction, reducing query response times by 50%.