AI Specialist

We are seeking a Speech AI Engineer to design, build, and deploy intelligent speech systems capable of converting human voice into accurate text and detecting emotional tone and intent in speech. 

Responsibilities:

• Design and develop speech-to-text (ASR) models using modern architectures such as Whisper, Wav2Vec2, or Conformer.
• Build speech emotion recognition (SER) models to classify tone, emotion, and mood from voice inputs.
• Collect, preprocess, and annotate custom speech datasets — dataset creation is part of this role.
• Apply data augmentation and noise-robust training for better real-world performance.
• Implement quantization, pruning, and optimization of models for real-time inference on servers.
• Develop and expose REST APIs for model access using FastAPI, ensuring scalability and security.
• Integrate speech models with text-processing and chatbot systems for unified voice–text experiences.
• Manage the full ML lifecycle — training, validation, deployment, monitoring, and continuous improvement.
• Containerize and deploy models using Docker and orchestrate services via Kubernetes.
• Continuously explore state-of-the-art speech and multimodal AI research to enhance accuracy and latency. 

Required Skills & Experience:

• Hands-on experience in speech recognition, speech emotion detection, or speaker identification.
• Proficiency in Python and deep learning frameworks:
• PyTorch, TensorFlow, HuggingFace Transformers
• Strong understanding of audio feature extraction (MFCC, log-mel spectrograms, CTC loss, etc.).
• Experience in model quantization and optimization (ONNX, TensorRT, TorchScript).
• Proficiency with FastAPI, Docker, and Kubernetes for scalable deployment.
• Familiarity with CI/CD, MLOps, and model monitoring workflows.
• Knowledge of GPU acceleration and multi-model inference management.
• Familiarity with databases (MongoDB, PostgreSQL, Redis).
• Practical experience with real-time or streaming audio pipelines is a strong plus.