Home | Blue Altair | Icebreaker

Home | Blue Altair

With Blue Altair's digital transformation, we help identify hidden insights, produce recommendations, and automate manual activities to optimize business outcomes.

2 members

Contact info

Link

bluealtair.com

Members

Faizan Qazi

•

3rd+

R&D Software Engineer MTS-II | Automation Consultant

James Ray

•

3rd+

AI Product Manager leading GenAI research and products since 2017. My mission is to empower creators and spark inspiration through human-centric ML at scale. At Netflix I built Studio GenAI from 0-1, working with filmmakers, game devs, and AI researchers to train and fine-tune text, image, and video generation for film and games. I partner closely with creators - including Academy Award and Emmy winners/nominees - to pioneer artist-driven methods for GenAI control and customization. Through novel foundation model controls, along with fine tuning and conditioning, we maximize the potential for human art direction in GenAI. Hand-in-hand with artists and filmmakers, we are unlocking new modes of human expression. We do so in a way that elevates and honors human creativity - with controllability at the core. --- Before Netflix, I led media understanding at Twitter - recommending content that sparks engagement, inspiration and curiosity, with a key focus on trust and safety. We used multi-modal AI to holistically understand text, images, and video - closer to the way humans do. That allowed us to make more engaging recommendations, and significantly increase trust & safety detection against adversarial actors. I also led generative AI for content creation. In 2017, I led an eng team that productized one of the first interactive GANs (before the rise of diffusion models.) Outside the day job, I'm a technical advisor to AI startups in medical LLMs & brain-computer interfaces, and I present at ML conferences (e.g., on deep reinforcement learning). Selected past work: --GenAI Model Customization and Controllability: Novel fine tuning + foundation models - maximizing the expressiveness of generative AI to augment and honor human creativity. --Distributed ML training platforms --LLMs and multi-modal embeddings for trust & safety, recommending, ranking, search (computer vision, LLM, and graph embeddings, factorization machines, learning-to-rank) --Interpretable AI for silicon photonics --Interactive generative adversarial networks (iGAN) for AI-augmented creative design --Multimodal video understanding --Spatiotemporal deep learning for video classification and brain-computer interfaces --Foundational language model for a virtual agent (contemporary state of the art)