AIML - Machine Learning Research Intern in Speech Recognition / NLP, SII

Key Skills

Apache Kafka, Automatic Speech Recognition (ASR), DevOps, JavaScript, Large Language Models - LLMs, Natural Language Processing (NLP)

Job Description

The Siri Information Intelligence team is seeking a highly motivated machine learning research engineer with experience in Speech and NLP applications. We are diverse and passionate group of engineers dedicated to bringing the best intelligent assistant experience, especially by tackling the issue of multi-linguality. The research intern will conduct cutting edge research to advance some of the most challenging and exciting problems of his/her own choosing within the overall scope as laid out here, with close supervision and support from our expert engineers, and will engage with the academic community by publishing innovative research and speaking at conferences and events.

Key Qualifications

• Strong background in applied machine learning and deep learning with experience in one of the following areas:
• Speech recognition including but not limited to end-to-end ASR architectures, language modelling, multilingual ASR, decoding algorithms and personalization and adaptation
• Natural Language Processing including but not limited to conversational interaction, machine translation, Knowledge Graph, Information Retrieval and Topic Modelling
• Foundation Models including but not limited to building infrastructure, datasets, model architecture and evaluation methodologies.
• Strong software engineering abilities to turn research ideas into near production level demo prototypes
• Have strong passion for new emerging research and technologies
• Publication records demonstrating innovative research is a big plus

Description

• This role will collaborate closely with engineers at Siri Language Engineering teams to address one of the following topics depending on the intern's expertise and interest:
• ASR: Code-switching and Mixed-language ASR, Subword NNLM, Application of LLM to speech recognition

• Multi-lingual foundation model with focus on lower resource languages: transfer learning from higher resource languages, data augmentation such as synthetic data generation, cross-lingual embeddings, etc. Also cultural sensitivity and reliability can be another research area.

• Data annotation leveraging large language model: research novel and creative methods to automate or assist the annotation process, making it faster, more consistent and more cost-effective.

Education & Experience

Currently pursuing a PhD, Master’s degree in Machine Learning, Speech Recognition, Natural Language Processing or related areas. PhD candidate preferred.