Indian Language AI Engineer (Speech, OCR & Translation)

About the role

Our chat, voice and document tools are used in English, Hindi and Bangla. You will improve how they listen, read and translate: speech that mixes languages, bills and notices photographed on a phone, and answers that read naturally in the user's language.

Make Cehpoint AI understand India as it speaks and writes: Hindi and Bangla speech, handwritten and printed documents, and translation that keeps meaning.

What you will do

• Build and evaluate speech-to-text and text-to-speech pipelines for Hindi and Bangla, including code-mixed speech.
• Improve OCR for printed and handwritten Indian documents such as bills, notices and forms.
• Create small, honest test sets in Indian languages and measure accuracy before and after every change.
• Work with the product team to ship improvements into chat, voice and the free AI tools.

What you should know

• Strong Python and practical experience with at least one of ASR, OCR or translation models.
• How Indic scripts are encoded and normalised (Unicode, conjuncts, numerals).

What you should be able to do

• Read and write at least one Indian language besides English.
• Turn a messy real-world sample into a reproducible test case.

How you work

• You notice when a translation is technically right but sounds wrong to a native speaker.

Skills we check

• Python
• Speech-to-text (ASR)
• OCR
• Machine translation
• Indic scripts and Unicode
• Evaluation datasets
• PyTorch

What you get

• Work on language problems most products ignore.
• Your improvements reach users in Hindi and Bangla directly.