Developed an agent-based workflow for extracting business requirements from source code, enabling users to efficiently generate a high-level overview of application modules. • Implemented a graph loader to convert codebases into graph representations, integrating an Gen AI chat interface to facilitate user interaction within the migration tool. • Developed a dependency mapping algorithm for EPAM’s GenAI Copilot that identifies dependencies in source code snippets and aligns them with the migration map, providing enhanced context for LLM and improving the migration process. • Conducted research on top-tier LLM models from various providers, enhancing the migration workflow by selecting the optimal model for a specific task improving overall efficiency and effectiveness. • Enhanced Retrieval-Augmented Generation (RAG) implementation for project documentation, enabling users to interact with a chat interface to ask specific questions about the project, improving accessibility and accuracy of information retrieval. • Improved Antlr4 semantic trees and grammars for supported languages, by incorporating the latest language features, achieving a substantial 30% improvement in parsing precision.