ML Engineer — Speech & NLP

CNCT CALL; Hyderabad; On-site; INR 1,400,000 - 2,200,000; Full time

Job summary

CNCT CALL in Hyderabad, India, seeks an NLP/Speech engineer to own STT/TTS quality for Indian languages, focusing on latency, code-switching, and production-grade audio pipelines on our voice platform.

You will fine-tune and evaluate ASR/TTS for Hindi, Telugu, Tamil, and English code-mix, build offline eval sets, run regression tests for model releases, and collaborate on streaming inference at 8-16 kHz.

Qualifications

• BE/BTech or MTech in CS/EE with 2+ years in speech or NLP.
• Comfortable with Python, PyTorch or similar, and shipping models to production.

Responsibilities

• Fine-tune and evaluate ASR/TTS for Hindi, Telugu, Tamil, and English code-mix.
• Build offline eval sets and regression tests for model releases.
• Collaborate with backend on streaming inference at 8-16 kHz.
• Document model cards, failure modes, and language-specific tuning.

Skills

Python

PyTorch

Model deployment

Education

B.E./B.Tech or M.Tech in CS/EE

Job description

The role

Own STT/TTS quality for Indian languages - latency, code-switching, and telephony-grade audio pipelines on our voice platform.

Who should apply

B.E./B.Tech or M.Tech in CS/EE with 2+ years in speech or NLP. Comfortable with Python, PyTorch or similar, and shipping models to production.

Expectations

• Fine-tune and evaluate ASR/TTS for Hindi, Telugu, Tamil, and English code-mix
• Build offline eval sets and regression tests for model releases
• Collaborate with backend on streaming inference at 8-16 kHz
• Document model cards, failure modes, and language-specific tuning