Xavier (Xavi) Amatriain

Curai Health is an end-to-end primary care service that combines the best medical team with state-of-the art AI as another member of the care team. I started the company from zero with my cofounder Neal Khosla, and two other founding engineers. As a founder, I did everything from figuring product and company strategy to writing AI research papers.

Advising SCRM - Lidl Digital International on product, organizational and technical issues

Leading strategy and product roadmap for the AI and Compute platforms that powers all of Google products. This includes many different tools and layers of the stack, namely: - AI Data: from human data annotation for pre and post-training to model and data lineage for GenAI data governance and RAI - AI Model activation: tools and processes for model pre-training, fine-tuning, prompt engineering, orchestration, and model serving at scale. While we mostly focus on Gemini, we also support many other workstreams for models like Imagen, Veo, or Gemma. In fact, we also support non GenAI models internally. - AI Frameworks: My team manages internal and external roadmap and strategy for frameworks like JAX, Keras, TensorFlow, or XLA. This also includes AI for mobile platforms now presented under the umbrella of Google AI Edge - Compute infrastructure and efficiency: We own much of the internal AI Compute infrastructure and build efficiency tools to make sure we can bring the best out of our hardware and data centers. - AI capacity planning and tools: Similarly, we model and plan capacity needs for all our Google needs across TPUs, GPUs, and standard compute. Our team collaborates closely with many other teams at Google. From Google Deepmind on the research side, all the way to Google Cloud and our many product teams (Search, Ads, Youtube...)

Leading company-wide Generative AI efforts across product and AI teams. This involved all the way from building the first LLM-based product feature prototype to designing our internal end-to-end Generative AI platform.

I led the Engineering team at Quora, creating the best source for knowledge on the internet.

Founded and led the Algorithms Engineering (AE) team using the latest machine learning and algorithmic advances to improve personalization, search, recommendations, and other product features at Netflix. We were the team building the famous Netflix recommendations, which revolutionized not only the product but the industry as a whole. The AE is a team of researchers/engineers whose focus is delivering innovation to the Netflix product through experimentation in the algorithmic space. To learn more about what I did at Netflix, read my post in Netflix' techblog: http://techblog.netflix.com/2012/04/netflix-recommendations-beyond-5-stars.html, the feature article in Wired (www.wired.com/underwire/2013/08/qq_netflix-algorithm/), or my interview in InfoQ http://www.infoq.com/interviews/data-science-roles-netflix-xavier-amatriain.

Senior Researcher in the Barcelona research center of Telefonica I+D. Working on Machine Leaning, Recommender Systems, User Modeling, Data Mining, Web and related areas. During this time I published over 20 peer-reviewed publications, was General Chair of the ACM Recsys conference, and contributed to the creation of Telefonica Research as one of the first researchers to join the team.

Teaching Software Engineering and Information Retrieval.

Managed the CLAM project (www.clam-project.org) since its very inception. CLAM became a framework of reference in multimedia and audio around the world. It won the 2006 ACM Award to the Best Multimedia Open Source Software and was a featured project for the 2007 Google Summer of Code. The CLAM project spurred the creation of IMMSound (http://en.wikipedia.org/wiki/Imm_sound), later acquired by Dolby.

Took the advanced track of the first edition of the ai-class online course by Sebastian Thrun and Peter Norvig. Note: This is not a course that is officially recognized by Stanford University and does not provide an official degree.

Took the advanced track of the first edition of the ml-class online course by Andrew Ng. Note: This is not a course that is officially recognized by Stanford University and does not provide an official degree.