In an industry facing hardware shortages and supply chain issues, Clika, a startup founded by Ben Asaf and Nayul Kim, is making strides with its AI compression platform. Clika aims to simplify the deployment of AI models into production by providing a toolkit that automatically “downsizes” internally developed models, reducing compute power consumption and improving inferencing speed. With techniques like quantization at its core, Clika’s compression engine utilizes an AI-based approach to understand and optimize unique AI model structures for the best performance.
Key Takeaway
Clika, a startup founded by Ben Asaf and Nayul Kim, is revolutionizing AI model compression with its platform. By providing a toolkit that automatically “downsizes” AI models, Clika enables companies to reduce compute power consumption and improve inferencing speed. With its AI-based compression engine, Clika distinguishes itself from competitors by tailoring compression methods to the unique structures of different AI models. As hardware shortages continue to impact the industry, Clika’s platform offers a timely solution to optimize existing resources and overcome supply chain constraints.
The Inspiration and Vision Behind Clika
Ben Asaf, co-founder of Clika, drew inspiration from his experience at Mobileye and his work on accelerating AI model training at Hebrew University. Recognizing the lack of industrial experience in implementing MLOps into AI development pipelines, Asaf sought to make AI more accessible, lightweight, and cost-effective for productization and commercialization. Together with Nayul Kim, a digital transformation consultant with enterprise-level expertise, they co-founded Clika in 2021.
Making AI “Tiny”, Faster, and More Affordable
At Clika, the process is simple. Companies can connect their pre-trained AI models and receive a compressed model that is fully compatible with their target device, whether it’s a server, the cloud, the edge, or an embedded device. Clika’s compression engine applies various methods, including quantization, to reduce the number of bits needed to represent information in a model without compromising its performance. This enables companies to achieve faster inferencing while also optimizing compute resources.
A Technological Advantage
Clika distinguishes itself from other startups in the AI compression space by leveraging its AI-based compression engine. While other solutions rely on rule-based techniques, Clika’s engine understands the unique structures of different AI models and applies the most suitable compression method for each one. Asaf claims that Clika has the world’s best compression toolkit for vision AI, surpassing existing solutions from industry giants like Meta and Nvidia.
Addressing Industry Challenges
The need for efficient AI models is growing as the industry grapples with hardware shortages. For example, Microsoft recently warned of potential service disruptions due to AI hardware shortages, and Nvidia’s highly sought-after AI chips are reportedly sold out until 2024. Clika’s platform provides a timely solution to these issues by offering AI model compression, enabling companies to make the most of their existing hardware and avoid supply chain constraints.
Rising Prominence and Future Plans
While Clika is currently in closed beta and pre-revenue, the startup has already secured $1.1 million in a pre-seed round with backing from investors such as Kimsiga Lab, Dodam Ventures, D-Camp, and angel investor Lee Sanghee. As the closed beta progresses, Clika plans to pursue seed funding in the near future. With its technological advantage and focus on addressing industry challenges, Clika is poised to make a significant impact on the AI compression landscape.
