Description

You will design, develop, and deploy Large Audio Language Models (LALMs) and Large Audio Reasoning Models for complex chain-of-thought tasks over speech and audio inputs.

Responsibilities

• Build Speech-to-Speech (S2S) systems including dialogue management and synthesis components.
• Research and implement alignment mechanisms between speech encoders and LLM backbones using adapters, LoRA, and efficient fine-tuning.
• Design efficient speech tokenization and temporal compression techniques for long-form audio reasoning.
• Optimize inference pipelines for low-latency, streaming applications in production.

Required Skills

• Master's degree in Computer Science, Electrical Engineering, or related field.
• 2+ years of industry or applied research experience in speech/audio AI, LLMs, or multimodal systems.
• Strong proficiency in Python and PyTorch with GPU-accelerated training experience.
• Solid understanding of speech signal processing, acoustic modeling, and audio representations.
• Knowledge of modern LLM architectures (Transformers, SSMs) and training paradigms like instruction tuning.
• Familiarity with modality alignment techniques such as adapter-based integration or cross-modal attention.

Preferred Skills

• Hands-on experience with speech frameworks such as ESPnet, SpeechBrain, or NVIDIA NeMo.
• Experience with distributed training (FSDP, DeepSpeed) and inference optimization (ONNX, TensorRT).

Key Skills

Education

Master's degree