Ashwin Swaminathan

I build at the edge of what’s possible with multimodal AI, computer vision, and generative models. My career has been about taking ideas from zero-to-one —turning ambitious concepts into real, scalable products that redefine how people and businesses interact with technology.

Currently, I am a Director of Applied Science at Amazon. I lead multimodal understanding efforts at Amazon, where I’ve helped shape and launch the Nova model suite—Nova Lite, Pro, and Premier. These models were designed to meet very different needs:

• Nova Lite brings ultra low-cost, lightning-fast multimodal inference for image, video, and text.

• Nova Pro strikes the balance between accuracy, speed, and cost—perfect for enterprise applications.

• Nova Premier is our most advanced model, built for complex, high-stakes tasks.

Alongside the core models, I’ve driven Amazon’s generative creativity initiatives. I helped bring to life:

• Nova Canvas, an image generation and editing model that now powers Virtual Try-On, and 

• Nova Reel, a text-to-video system that expanded into two-minute videos and advanced storyboarding with its 1.1 release. 

• I also led the launch of the Titan image generator, pushing state-of-the-art performance.

My focus has always been not just on building powerful models, but on positioning them responsibly—ensuring Nova became a competitive multimodal platform that businesses and creators can adopt with confidence and scale.

Before Amazon, I served as Senior Director of Perception at Magic Leap, leading the World Sensing team that developed the foundation for spatial computing on ML2. My team worked on SLAM, world reconstruction, object recognition, and scene understanding—everything required for devices to understand and interact with the physical world. I also helped architect and launch Magic Leap’s AR Cloud, enabling persistence, multi-user sharing, large-scale mapping, and world understanding across devices. Some of our milestones at Magic Leap included:

• Persistence and sharing in the MagicVerse SDK (2019)

• XRKit for iOS and Android (2020)

• 3D object recognition with bounding boxes (2020)

Even earlier, I was a Senior Staff Engineer/Manager at Qualcomm Research, where I worked on computer vision and machine learning algorithms for Snapdragon chipsets, powering use cases like augmented reality, virtual reality, and drones. At Qualcomm, I also led several academic collaborations and served on the Qualcomm patent review board, helping shape innovation strategy.

My journey began in academia. I earned my Bachelors in Electrical Engineering from the Indian Institute of Technology, Madras, and my M.S. and Ph.D. in Electrical and Computer Engineering from the University of Maryland, College Park. My doctoral research focused on multimedia forensics, exploring intrinsic and extrinsic fingerprints for information security. During my Ph.D., I interned at HP Labs (2006) on digital rights management and at Microsoft Research (2007) on machine learning and applied statistics.

Across Amazon, Magic Leap, Qualcomm, and academia, a common thread runs through my work: building systems that give machines new ways to perceive, understand, and create. Looking ahead, I’m excited to continue pushing the boundaries of multimodal AI and to collaborate with others who share that vision.

If you’re passionate about foundational models, generative AI, deep learning, or computer vision, let’s connect—we’re hiring, and I’d love to hear from you.