Sree Sai Bhavesh Veerabomma

• Designed multi-agent systems to automate complex workflows in banking sector reducing manual worload by 40% by enabling task-specific decision-making, and guide user journeys with dynamic prompt templates and tool-based flows. • Built a custom semantic RAG pipeline using LangChain, Elasticsearch and FastAPI achieving a 50% improvement in document response accuracy through dynamic knowledge retrieval from documents or data sources. • Reduced LLM API costs by 30% using semantic caching techniques • Used Ragas, DeepEval tools for rigorous model evaluation and implemented different (10) metrics to evaluate model, rag and agentic systems wrt their behavior and performance. • Developed a synthetic dataset generation pipeline to create diverse, traceable Q&A pairs for training and evaluating document-level question-answering models. • Led incremental model development to dynamically identify the correct document for a given user query and role, significantly boosting system precision in document retrieval. • Contributed to full-stack development by designing and implementing UI components using React & TypeScript, and seamlessly integrating with backend services built on FastAPI (for complex workflows like knowledge processing), & nest.js for dynamic data interaction and workflow management

• Gathered, reviewed, analyzed and stored data performing Google Search Scraping. • Scraped 30K+ real estate listings across 210 cities using Python and AWS automation. • Built dashboards to analyze trends and optimize marketing decisions

Machine Learning, AI, Natural Language Processing, Databases, Computer networks