Company attributes
Other attributes
Latent AI is a developer of software for machine learning (ML) and artificial intelligence (AI) solutions for edge computing deployments. It has developed a platform to train, quantize, and deploy edge AI networks. The company offers a quantization optimizer for edge AI devices, which can automate the exploration of low-bit-precision AI training in neural networks and for on-device intelligence and inference. The company claims to enable optimization for computing, energy, and memory without requiring changes to existing AI or ML infrastructure or frameworks.
Latent AI was founded in 2018 by Jags Kandasamy and Sek Chai and was spun out of SRI International in 2019 as a technology spinout with the Latent AI Efficient Inference Platform (LEIP). The platform is based on years of DARPA research and is developed to empower developers and their edge AI projects to optimize within compute, energy, and memory budgets. The company is based in Menlo Park, California.
The Latent AI Efficient Inference Platform (LEIP) is a software tool developed to bring AI models to devices with reliability and security and is dedicated to edge MLOps workflows. The machine-learning-operations software-development-kit (MLOps SDK) is developed to help users produce efficient, compressed, and secured models for compute-constrained devices. The optimization is done without requiring changes to existing AI or ML infrastructure and frameworks. The LEIP platform offers developers customizable templates that Latent AI calls Recipes, which are pre-qualified to given hardware with maintenance and configuration reduced to a single command line call.
The platform is designed to be scalable and seamless, integrating into a user's trusted CI/CD software development flow. It is built to be secure with unique identifiers and integrity checks to prevent tampering. It has already been selected by Booz Allen Hamilton to support the Department of Defense's Chief Digital and Artificial Intelligence Office with deploying machine learning models at the tactical edge. This is based on the needs of the DoD to deploy AI and ML models into small form factor chipsets on tactical devices.
