TeamIDE Labs
Applied AI research. Building inference, training, and systems infrastructure from the ground up.
Foundry
Pure C machine learning framework. Zero external dependencies, full tensor operations, quantization, and training. Learn more →
Inference Engines
Two approaches to running AI locally. A universal model server and a bare-metal boot environment. Learn more →
Barswap
GPU multiplexer for dense compute systems. Unlock all your GPU hardware on commodity motherboards. Learn more →
Distributed Training
Training language model adapters across commodity laptops with zero inter-node communication. Published research on the Research page.
Memory Systems
Extended context and retrieval systems for language models. Multi-tier memory architectures that keep RAM bounded while enabling long-horizon reasoning.
AI-Native OS
Operating systems that boot directly into AI inference. From UEFI bare metal to full runtime. No Python, no Docker, no cloud.
Model Architecture
Novel hybrid architectures combining recurrent and attention mechanisms. Exploring how architectural choices affect hardware utilization and memory efficiency.
Read our published work on the Research page.