Applied Scientist II, Alexa International Tech
Core
Develop novel algorithms and modeling techniques for Large Language Models (LLMs), ASR, TTS, and Speech-to-Speech models to advance Alexa's international products and services.
Role type
Senior IC Applied Scientist (Generative AI & Speech)
Builds
Multimodal personal digital assistants leveraging LLMs, ASR, TTS, and Speech-to-Speech models
Domain
Consumer Technology / Speech & Language AI
Deliverable
production ML models
Required skills
Deep learning, generative models, LLM architectures, ASR/TTS architectures, end-to-end speech models, LLM evaluation & tooling, SFT, DPO, RLHF, RLAIF, numerical optimization, parallel and distributed computing, high-performance computing
Preferred skills
State-of-the-art LLM training and post-training techniques, Unix/Linux environment usage
Technologies
Python, Java, C++, LLM frameworks, ASR/TTS frameworks
Responsibilities
Analyze and model customer behavior based on large-scale data; Build novel online & offline evaluation metrics for multimodal assistants; Drive research in ASR, TTS, and Speech-to-Speech model training and fine-tuning; Advance multilingual speech recognition and synthesis using LLM-based architectures; Fine-tune/post-train LLMs using SFT, DPO, RLHF, and RLAIF; Collaborate with partner teams on evaluation frameworks and post-training methodologies; Contribute to the scientific community through publications and community engagement
