Sqwish


Programme

Pre-Accelerate Cambridge

Sector

Computing/Data/Software


Sqwish is building super tiny and super fast models that make AI workflows more efficient for businesses. They are launching with a prompt compression API that reduces the size of input to AI models, lowering response time and costs, without sacrificing response quality.

Founding team

Ushnish Sengupta

Ushnish successfully defended his PhD in Engineering from the University of Cambridge in 2023, specializing in applications of Bayesian machine learning and optimization.

He co-founded IDEOM AI, focusing on LLM-powered creative thinking tools. His experience includes senior AI research at Mediatek and internships at Rolls-Royce and the German Aerospace Center, applying machine learning to aerospace engineering problems.

Ushnish has published extensively, including papers at NeurIPS and the Journal of Chemical Physics, and has received accolades such as the Marie Sklodowska-Curie fellowship.

Federica Freddi

Federica holds a Master of Engineering from the University of Cambridge and has a robust background in AI and software development.

She co-founded IDEOM AI and served as CTO, leading full stack development and user research.

At Mediatek Research, she contributed to deep learning projects, focusing on meta-learning and neural network optimization. Federica has also interned at ARM and Bending Spoons, working on data engineering and iOS development, respectively.

Her achievements include co-authoring the FishLeg optimizer, spotlighted at ICLR, and growing a social media account to 200k followers using an AI-driven automation tool.

Top