Alexis Conneau Email & Phone Number | I am an OpenAI researcher focused on Multimodal and Audio AGI. I've led the research towards "Her" (GPT4-o and GPT-5), the first GPT models natively integrating Audio. I have 10 years of experience in deep learning for natural language processing. My journey includes significant contributions at OpenAI, Facebook AI Research and Google, in text and speech neural networks. My key achievements include: - Creating the first version of "Her" at OpenAI, through GPT-4o. I brought that vision to OpenAI and built that Audio LLM research (almost) from scratch in duo with a colleague. - Creating the first neural networks capable of translating text and transcribing speech without any supervision (Unsupervised MT and ASR). - Building universal multilingual masked language model representations for text and speech (XLM-R, Wav2vec 2.0 XLS-R). - Shipping these language model representations at Meta and Google, impacting over 1 billion users across various applications such as text classification, machine translation, and speech recognition. - Finding the concept of 'zero-shot cross-lingual transfer', demonstrating that a multilingual language model, once fine-tuned on a single language, can effectively operate across other languages (XNLI, XLM) - Open-sourcing Facebook AI Research's most popular foundational text and speech masked language models (XLM-R and Wav2vec 2.0 XLS-R), available on Hugging Face (https://huggingface.co/models?sort=downloads). I have been recognized with best paper awards at EMNLP (2017, 2018) and SLT (2022), with publications across top journals and conferences like JMLR, NeurIPS, and ICML. My work has significantly enhanced text and speech systems for hundreds of millions, featured in the Washington Post: https://www.washingtonpost.com/technology/2021/07/01/humans-ai-language-advances-risks/

Alexis Conneau

I am an OpenAI researcher focused on Multimodal and Audio AGI. I've led the research towards "Her" (...
San Francisco Bay Area, San Francisco Bay Area, United States

*50 free lookup(s) per month.

No credit card required.

Alexis Conneau’s Email alexis.conneau@openai.com

Alexis Conneau’s Phone Numbers

Get ALEXIS’s Phone Number (For Free)

Social Media

Alexis Conneau’s Location San Francisco Bay Area, San Francisco Bay Area, United States

Alexis Conneau’s Current Industry Openai

Alexis Conneau’s Prior Industry Meta | Google | Ecole Normale Superieure Paris Saclay

Not the Alexis Conneau you were looking for?

Find accurate emails & phone numbers for over 700M professionals.

Work Experience

Openai

Member of Technical Staff

2023-04-01 00:00:00 - Present

Google

Research Scientist

2021-04-01 00:00:00 - 2022-04-01 00:00:00

Ecole Normale Superieure Paris Saclay

Lecturer (Master MVA) - Deep Learning for NLP

2015-10-01 00:00:00 - 2019-03-01 00:00:00

Skills

No Skills Available

Languages

No Languages Available

FAQ About Alexis Conneau

Alexis Conneau is located at
San Francisco Bay Area, San Francisco Bay Area, United States

Alexis Conneau is a I am an OpenAI researcher focused on Multimodal and Audio AGI. I've led the research towards "Her" (GPT4-o and GPT-5), the first GPT models natively integrating Audio. I have 10 years of experience in deep learning for natural language processing. My journey includes significant contributions at OpenAI, Facebook AI Research and Google, in text and speech neural networks. My key achievements include: - Creating the first version of "Her" at OpenAI, through GPT-4o. I brought that vision to OpenAI and built that Audio LLM research (almost) from scratch in duo with a colleague. - Creating the first neural networks capable of translating text and transcribing speech without any supervision (Unsupervised MT and ASR). - Building universal multilingual masked language model representations for text and speech (XLM-R, Wav2vec 2.0 XLS-R). - Shipping these language model representations at Meta and Google, impacting over 1 billion users across various applications such as text classification, machine translation, and speech recognition. - Finding the concept of 'zero-shot cross-lingual transfer', demonstrating that a multilingual language model, once fine-tuned on a single language, can effectively operate across other languages (XNLI, XLM) - Open-sourcing Facebook AI Research's most popular foundational text and speech masked language models (XLM-R and Wav2vec 2.0 XLS-R), available on Hugging Face (https://huggingface.co/models?sort=downloads). I have been recognized with best paper awards at EMNLP (2017, 2018) and SLT (2022), with publications across top journals and conferences like JMLR, NeurIPS, and ICML. My work has significantly enhanced text and speech systems for hundreds of millions, featured in the Washington Post: https://www.washingtonpost.com/technology/2021/07/01/humans-ai-language-advances-risks/