VADE

An LLM inference engine, written in Rust, built for the DGX Spark.

• Rust from the server down to the GPU kernels
• No Python. No CUDA toolkit.
• Runs on one Spark, scales across many
• Coming soon