I’m a software engineer who works at the intersection of legacy systems and modern AI tooling. My day job involves migrating decades-old codebases to modern stacks — proprietary languages to Python 3, analysis of database layers, undocumented business logic, and the kind of technical debt that accumulates when a system runs production workloads for 40+ years.
I work with what’s already there.
What I Do
I build AI-assisted tooling for legacy code migration: translation pipelines, automated code review systems, parity testing frameworks. The systems I work with use proprietary languages that no LLM has seen in training, so most of the interesting problems are about making AI useful where it has zero prior knowledge.
Current stack: Python 3, Java, C, C++, llama.cpp, FAISS, LLM APIs, FastAPI, Docker, GitLab CI/CD, Oracle, PostgreSQL.
Background
Computational linguistics by education. Software engineering by profession. I started in CPQ (Configure-Price-Quote) implementations for enterprise customers, moved into legacy system modernization, and found my niche where AI meets code that predates the internet.
What I Write About
This blog documents what I learn while building AI tooling for real-world migration projects. No hype. What works, what doesn’t, and where the tools fall short.
Posts come in batches. I write after a project milestone has been completed. There is no point in sharing experiences if your next post is “Forget all I said before, I made error x and lost two weeks of progress.” Fun to read? Yes. Human? Yes. Any value in putting that online? No.
Contact
Find me on LinkedIn or reach out if your codebase gives you migraines and you’re looking for someone who’s been there.