Boosala Sainadh

DataHub LLMOps

• Designed and developed Generative AI and NLP solutions using Python, Transformers, GPT-4, Whisper, LangChain, and LlamaIndex to process large-scale unstructured data from audio, video, and document sources.
• Developed Video-to-Knowledge Converter (V2K) that transformed video recordings into structured knowledge documents using Whisper transcription and GPT-based knowledge extraction pipelines, enhancing accessibility of information.
• Built scalable Retrieval-Augmented Generation (RAG) pipelines using text chunking, embeddings, and Azure Cognitive Search vector databases enabling semantic and hybrid search across enterprise knowledge repositories.
• Optimized embedding generation, chunking strategies, and vector retrieval parameters, resulting in improved semantic search accuracy and reduced response latency for knowledge retrieval.
• Developed MeeMinutes, an AI-powered meeting intelligence system that generates summaries, action items, and insights from meeting transcripts and recordings using LLM-based multimodal processing.

RAG Natural Language to GraphQL Queries

• Designed and implemented a RAG-based conversational AI system that converts Natural Language Queries (NLQ) into GraphQL queries and natural language responses.
• Developed FastAPI and GraphQL APIs to integrate AI services with enterprise applications and backend data platforms.
• Implemented Agentic AI workflows using CrewAI with MCP tools and MCP server integrations, enabling dynamic retrieval and processing of enterprise data by AI agents.
• Deployed scalable AI solutions using Azure OpenAI, Azure Blob Storage, Azure Cosmos DB, Azure Functions, and Azure Web Apps with Docker containerization and CI/CD pipelines.