Shankar Ananthakrishnan

About Shankar Ananthakrishnan

Shankar Ananthakrishnan has 20+ years of hands-on and technical leadership experience in AI spanning core foundation model development for Amazon Nova, as well as Conversational AI applications such as Alexa.Shankar received his B.Eng. in Electronic & Telecommunication Engineering from the University of Mumbai, India, and his M.Sc. and Ph.D. degrees, both in Electrical Engineering, from the University of Southern California. He is currently a Director of Applied Science at Amazon AGI, leading pre-training, post-training, and AI Safety efforts for Amazon Nova foundation models. Previously, Shankar led a range of Natural Language Understanding (NLU) efforts for Amazon Alexa, and Machine Translation initiatives at Raytheon BBN Technologies.Shankar\'s research interests are broadly centered around foundation model development (pre- and post-training), speech recognition, statistical machine translation of text and speech, natural language understanding, and related areas in statistical pattern recognition and machine learning. He has published over 50 papers in peer-reviewed conferences and journals, and is the recipient of best paper awards at leading conferences, including ICASSP 2005 and Interspeech 2010.Specialties: foundation models & LLMs, speech recognition, machine translation, natural language understanding.Publications: & hl=en

Experience

1. Director, Applied Science (Amazon Agi)

   Amazon

   Apr 2021 — Present · Boston, MA, US

   June 2023 - Present: Defining and driving the R&D roadmap for Amazon\'s Nova family of multimodal foundation models. Reporting directly to the Amazon AGI SVP, I lead teams focused on large-scale pre-training, post-training and AI Safety for all Nova 1.0 models (Micro, Lite, Pro, Premier) and Nova 2.0 models (Lite, Pro, Omni). Supporting R&D in advanced foundation model topics such as architecture design, long-context capabilities, scaling science, and efficient model training for dense and mixture-of-expert models.April 2021 - May 2023: Led the vision and AI strategy for Alexa\'s language understanding technology. Managed a multi-functional applied research and development organization that invented and productized state-of-the-art deep learning techniques for large-scale NLU/NLP applications. Led the development and open-source release of AlexaTM 20B, a multilingual seq2seq LLM specialized in synthetic data generation and machine translation.

Education

• University of Southern California

  M.Sc., Electrical Engineering

• University of Southern California

  Ph.D., Electrical Engineering

• University of Mumbai

  B.Eng., Electronic & Telecommunication Engineering

Skills

• C++
• Machine Learning
• Perl
• Python
• Latex
• Computer Vision
• Natural Language Processing
• Matlab
• Programming
• Signal Processing
• Pattern Recognition
• Algorithms
• Statistical Machine Translation
• Speech Recognition
• Speech Processing