Website:
chariot.in
Job details:
Language Software Engineer
Full-time
Location: Delhi / Remote
About Chariot: Chariot is a frontier AI research lab based in India, building foundational AI models. We're one of four teams selected under the IndiaAI Mission to build models from India. We want our models to work as well in Indian languages as the best AI systems work in English.
The Role: You'll build the software layer that trains our models to work across languages from the text processing and data pipelines to the systems that measure whether it's any good.
This includes:
- Language infrastructure. The layer every model runs through: tokenization, Unicode and script handling, transliteration and romanization, and code-switching between languages like Hindi and English.
- Data pipelines. Multilingual text and audio arrives messy. You build the pipelines that collect, clean, filter, deduplicate, and align it into training-ready datasets, across dozens of languages and scripts.
- Evaluation. You build the harnesses that measure how well a model performs in each language, turning the test sets and error patterns our language researchers find into automated, repeatable evals.
- Working with research. You sit between the language researchers who know the languages and the ML researchers who train the models, incoparating what they find directly into how the models behave.
Qualifications & things we look for:
- Strong software engineering skills, especially in Python. You write code other people can build on.
- A real understanding of how language works in software: encodings, Unicode, scripts, tokenization, transliteration.
- Experience building data pipelines that process large volumes of text or audio.
- You know at least one Indic language well, or you're genuinely motivated to build for languages beyond English.
- Experience with speech (ASR or TTS), NLP, or multilingual systems is a strong plus.
- You're comfortable operating in an early, fast-moving environment and can own projects end to end.
- A degree in computer science or a related field, or equivalent practical experience.
How to apply: Send us your resume, a link to code or projects you'd want to point to, the language(s) you know or care about building for, and a short note on why this role interests you.
Click on Apply to know more.