AI engineer & translation-technology builder
AI engineer building LLM-powered translation systems, writing about agents, language models, and how meaning survives the move between languages.
Arjune Mirchandani is an AI engineer who builds with large language models — designing the systems, prompts, evaluations, and guardrails that turn a capable model into something dependable in production. His focus is translation: using LLMs to carry meaning between languages more naturally than earlier machine-translation approaches allowed, and being precise about where they still fall short.
That work shapes what he writes for TranslateThis. Translation is an unusually honest test of an AI system — the output is judged by people who know both languages, and a fluent-sounding mistranslation is worse than an obvious one. He covers translation, localization, and language tooling with the same standard he applies to his own systems: explain what a tool actually does, including its real limitations rather than just its strengths.
Arjune also writes about AI engineering more broadly — agent design, bounded autonomy, and the practical discipline of shipping software built on models that are powerful but not reliably predictable.
Underneath the AI work is 25 years of full-stack engineering: Next.js, React, Angular, TypeScript, and Node.js, plus the production instincts that come from shipping and maintaining real systems. He builds fast, accessible, SEO-friendly applications — TranslateThis itself is one, a statically generated Next.js site meant to load quickly and be genuinely useful.
You can find more of Arjune's work and connect with him through the links below.
Articles Arjune has published on other platforms.
Autonomy is a demo metric; bounded is a shipping strategy. Why unbounded overnight agent runs create verification debt faster than teams can review it — and the four properties that make a task agent-sized.
· Originally published at arjune.me
Read on Medium ↗