Deep Learning and AI technologies are disrupting every industry in an incredible way – self-driving cars, drones, surveillance cameras, detecting cancers, virtual assistants and much more. Under the hood of many exciting AI technologies are Deep Neural Networks (DNNs): powerful algorithms that deliver state-of-the-art performance on tasks such as recognizing objects in photos, translating speech and detecting patterns in video. However, DNNs are massively complex, which makes them inaccessible for use on everyday devices and incurs high costs in data centers. The recent emergence of deep learning and AI solutions on mobile and low-cost devices has been limited in performance and affordability, ultimately preventing people and businesses from using these devices.
To solve this, Deeplite created Neutrino, an intelligent optimizer that makes fast and highly-efficient AI algorithms for use in daily life. Our software enables AI to run on everyday devices like smartphones, cameras and vehicles which were previously not able to use sophisticated AI solutions. Deeplite’s patent-pending engine can automatically satisfy computation constraints for DNNs in challenging real-time and large-scale applications. The result is a robust and efficient AI for daily environments. We work to provide companies and developers robust and adaptive deep learning to unlock high-value opportunities for AI in edge computing, a new paradigm that can decrease latency, protect user privacy and do not rely on constant internet connectivity. Through energy-efficient DNNs, we are unlocking new opportunities previously not possible for deep learning, one optimization at a time
Based in Montreal since April 2018, our team consists of driven entrepreneurs, passionate researchers, experienced professors and advisors from around the globe who are determined to make AI more accessible and affordable. We have been developing AI technology over the years of university research at top universities in the US and Canada. The inventions produced at Deeplite, combined with the work of our partners in academia has enabled the most advanced and scalable deep learning available. To achieve the complex task of optimizing state-of-the-art DNNs, our solution uses an AI software agent to automate the process of designing DNN models. This intelligent agent has been trained through experience to “learn” the right DNN model for any task. The agent enables engineers to meet challenging computation constraints on edge devices for intensive tasks like recognizing objects, gestures, and video analytics. Deeplite helps our partners in the smartphone, automotive and IoT industries to deploy real-time and sophisticated deep learning at a cost-effective power, computing, and engineering budget. Our software aims to improve the profitability, scalability, and opportunities for real-world AI and deep learning tasks.
At Deeplite, our goal is to solve neural efficiency for AI computing. By tackling fundamental challenges with how deep learning will aid the future, we aim to make this powerful technology more efficient and accessible for the next generation of intelligent systems. To do so, we work with our partners in academia, our strategic advisors and talented team to not just predict the future of AI, but to make it.
Ehsan Saboori Co-founder, Technology Lead
Davis Sawyer Co-founder, Product Lead