AI/ML Engineer – Speech & TTS

real-time Voice & TTS systems

passionate AI/ML Engineer

Key Responsibilities:

TTS Development & Optimization

• Design, train, and fine-tune text-to-speech (TTS) models

  with

  very low inference latency

  suitable for real-time applications.
• Work on

  end-to-end pipelines

  – text normalization, phoneme conversion, acoustic modeling, vocoder optimization.
• Implement

  streaming & low-latency architectures

  for speech generation.

Multilingual Speech Systems

• Build scalable TTS engines supporting all major Indian languages

  .
• Handle

  code-mixing & transliteration

  (e.g., Hinglish, Tanglish).
• Train

  multilingual or cross-lingual TTS models

  leveraging transfer learning.

Model Training & Deployment

• Collect, clean, and preprocess large-scale Indian language speech datasets

  .
• Train models using frameworks like

  PyTorch, TensorFlow, or JAX

  .
• Optimize for

  GPU/TPU inference

  with libraries like

  ONNX, TensorRT, OpenVINO

  .
• Deploy models in

  real-time APIs

  (FastAPI, gRPC, WebRTC, LiveKit, etc.).

Research & Innovation

• Experiment with latest speech synthesis architectures

  (FastSpeech2, VITS, Glow-TTS, HiFi-GAN, etc.).
• Improve

  voice naturalness, prosody, and expressiveness

  while keeping

  latency <150ms

  .
• Explore

  few-shot or zero-shot voice cloning

  for Indian voices.

Required Skills & Qualifications:

• Strong background in

  AI/ML, Speech Processing, or Computational Linguistics

  .
• Hands-on experience with

  TTS frameworks

  (Tacotron, FastSpeech, VITS, Glow-TTS, StyleTTS2, Vall-E, etc.).
• Expertise in

  deep learning frameworks

  (PyTorch, TensorFlow, JAX).
• Experience in

  low-latency inference optimization

  (ONNX, TensorRT, quantization, pruning).
• Knowledge of

  Indian languages, phonetics, and linguistic diversity

  .
• Experience with

  large dataset handling, audio preprocessing, feature extraction (MFCC, mel-spectrograms)

  .
• Familiarity with

  streaming systems

  and

  real-time audio APIs

  .
• Strong programming skills in

  Python, C++ (for inference optimization), and cloud deployment (GCP/AWS/Azure)

  .

Preferred / Nice-to-Have

• Research publications or contributions in

  speech synthesis / TTS / ASR

  .
• Experience with

  speech data augmentation

  techniques.
• Prior work on

  cross-lingual speech models

  or

  Indian-language AI projects

  .
• Familiarity with

  Rust/Go

  for real-time systems.
• Experience with

  Voice Cloning, Emotion Modeling, and Prosody Transfer

  .

What We Offer

• Opportunity to

  build India’s next-generation speech AI platform

  .
• Work on

  challenging problems

  in

  real-time multilingual AI voice systems

  .
• Competitive compensation with

  equity options

  (if applicable).
• A fast-paced, collaborative startup culture with

  direct impact on product innovation

  .