Spearheading AI-Driven Legacy Modernization at Ippon Technologies
Ippon Technologies is a consulting firm specializing in digital transformation, cloud, and modern software engineering for large-scale clients. As a Full-Stack Software Engineer since November 2025, I lead internal R&D initiatives aimed at accelerating legacy modernization — turning slow, manual migration processes into automated, AI-assisted workflows.
AI-Driven Modernization CLI in Rust
I designed and built a command-line tool in Rust that leverages Tree-sitter for Abstract Syntax Tree (AST) parsing of legacy codebases. Combined with a Retrieval-Augmented Generation (RAG) pipeline, the tool automatically generates structured documentation from existing code — enabling engineers to understand complex, undocumented systems in hours rather than weeks.
The choice of Rust was deliberate: memory safety, near-zero runtime overhead, and native compilation make it ideal for processing large codebases quickly. Tree-sitter provides language-agnostic parsing, allowing the tool to support multiple legacy languages from a unified architecture.
Agentic Workflows with Model Context Protocol (MCP)
I implemented agentic workflows using the Model Context Protocol (MCP), creating autonomous testing agents and refactoring assistants that interact directly with IDEs and file systems. These agents can analyze code, propose refactors, run test suites, and report results — reducing manual overhead on modernization tasks significantly.
The MCP-based approach enables context-aware agents that understand the full project structure, making interventions more precise than traditional static analysis tools.
Developer Experience & “Chat with Code”
A key outcome of these initiatives was a "Chat with Code" capability — an interactive interface allowing engineers joining legacy projects to query the codebase in natural language. This dramatically reduced onboarding time on complex systems, giving new team members a productive starting point from day one.
Technical Stack
Rust · Tree-sitter · RAG (Retrieval-Augmented Generation) · Model Context Protocol (MCP) · TypeScript · Node.js · LLM APIs