Our Mission

Pioneering Accessible AI Research.

We are decentralizing intelligence by focusing on offline models and small language models (SLMs) that empower individuals and local institutions, reducing dependency on big tech infrastructure.

Scientific Representation of AI Research
98%
Efficiency increase in edge-deployed SLMs.

Research Pillars

wifi_off

Offline-First Architectures

Developing models that operate entirely without internet connectivity, ensuring data sovereignty and reliability in remote environments.

Learn More arrow_forward
compress

Small Language Models

Optimizing massive LLMs into agile, highly efficient Small Language Models (SLMs) that run on local consumer hardware.

Embedded AI Hardware
memory

Embedded AI

Our research bridges the gap between software optimization and hardware constraints, tailoring AI for the next generation of RISC-V and ARM edge devices.

tune

AI Optimization

Advancing techniques like quantization, knowledge distillation, and efficient fine-tuning to run powerful models on resource-constrained devices without sacrificing accuracy.

Explore Techniques arrow_forward

Rigorous Inquiry, Real Impact

Our multidisciplinary team combines academic depth with pragmatic engineering to solve the AI accessibility gap.

Quantization Strategies
Archive 2024 SLM

Quantization Strategies for Edge LLMs

Evaluating 4-bit and 2-bit quantization on low-power ARM architectures for real-time natural language processing tasks.

Download PDF download
Decentralized Intelligence
Ethics Societal

Decentralized Intelligence & Data Privacy

A study on how local model deployment fundamentally shifts the power dynamic between users and monolithic tech corporations.

Download PDF download
Democratizing AI
Education Accessibility

Democratizing AI Literacy through SLMs

How affordable, offline models are transforming educational outcomes in bandwidth-limited regions around the globe.

Download PDF download

Become a Research Associate

Join our multidisciplinary team of scientists and engineers. Apply to work on cutting-edge offline models, embedded AI, and ethical tech.