About the job
Join us in creating a safer world, one incident at a time.
At Ambient.ai, we are at the forefront of AI-driven physical security, empowering leading enterprises to mitigate risks, enhance operational efficiency, and derive actionable insights. Our innovative platform is trusted by seven of the top 10 technology companies in the U.S. and numerous Fortune 500 firms, revolutionizing the way they approach physical security.
Our cutting-edge solution integrates seamlessly with existing camera and sensor systems, utilizing advanced AI and computer vision to enable real-time monitoring and proactive threat detection. By significantly reducing false alarms by over 95%, Ambient.ai allows security teams to concentrate on genuine threats and avert incidents before they arise.
Founded in 2017 and supported by renowned investors such as Andreessen Horowitz, Y Combinator, and Allegion Ventures, Ambient.ai is a Series B company dedicated to making every security incident preventable.
Interested in learning more? Connect with us on LinkedIn and YouTube
About the Position:
Ambient.ai is seeking an innovative Applied Research Scientist to lead the development of next-gen foundation models for computer vision. In this pivotal role, you will collaborate with a talented team to create multimodal models that achieve state-of-the-art performance on real-world vision benchmarks. You will take charge of the entire model development cycle, from pre-training and fine-tuning on image-language datasets to deploying models through distillation and compression techniques. This hands-on, cross-functional role offers a unique opportunity to directly influence our mission of preventing security incidents.
Key Responsibilities:
Develop & Optimize VLMs: Create and enhance transformer-based vision-language models that comprehend images, videos, and text, optimizing for real-time inference.
Pre-training & Fine-tuning: Manage the complete training pipeline—from pre-training on image-text data to fine-tuning tailored for Ambient.ai’s physical security context and applications.
Model Compression & Optimization: Implement techniques such as distillation, quantization, and pruning to minimize model size and latency, facilitating efficient edge deployment.
Leverage Open-Source & Innovate: Utilize and enhance leading open-source models. Prototype novel architectures and training methodologies to propel Ambient.ai’s multimodal AI research forward.
Collaborate: Work closely with cross-functional teams to integrate models into production and ensure alignment with our strategic objectives.

