Ippon Technologies
Designed and built an internal AI platform for legacy modernization at Ippon Technologies: a Rust CLI, a web app and an MCP server on one core, rolled out to 120+ consultants.
- ~600 employees
- Legacy codebase
- Hybrid
- Language
- RustTypeScript
- AI/ML
- RAGLLM APIsTree-sitter
- Protocol
- MCP
- Runtime
- Node.js
An internal AI platform for legacy modernization
Ippon Technologies is a consulting firm of about 600 people working on cloud and software engineering for large clients. Since November 2025 I have split my time between client delivery and internal R&D. The R&D side produced one product: a platform that helps engineers understand and modernize legacy codebases.
What I built
I designed and built the whole product. It has three surfaces that share one core:
- A Rust CLI that parses codebases with tree-sitter, chunks them along the syntax tree and indexes them for retrieval.
- A web app where engineers index their projects, read the generated documentation, explore a graph of the code and ask questions about it.
- An MCP server that gives coding agents and IDE assistants the same index, for exploration, testing and refactoring tasks.
Around that core sit the parts that make it a product rather than a demo: separate workspaces per organization, user and access management, logs, and tracking of LLM usage and cost per organization.
The core is written in Rust so that one engine serves the CLI, the web app and the MCP server, and so that parsing large repositories stays fast. Tree-sitter keeps the parsing language-agnostic, which matters when the codebases in question span several legacy languages.
Where it stands
The platform is rolled out internally to 120+ consultants.
Client delivery
Alongside the platform I work on client engagements, currently Suez IWS: a regulated B2B extranet for hazardous-waste tracking, where I rebuilt the CI/CD pipeline and moved a legacy Liferay front end to Angular micro frontends.
Writing
I also publish technical articles on the Ippon Tech Blog, on topics explored during this R&D work.