Bobby Mathews Blog

Rust Systems Engineer turning volatile compute into reliable capacity

About

Independent consultant specializing in distributed systems, high-performance computing, and GPU programming. I build infrastructure that demands correctness and performance.

Currently building toward autonomous infrastructure: systems that are self-aware, self-healing, and self-governing. Starting with LLM inference orchestration, then expanding horizontally.

Philosophy

Long-term vision: Applications that deploy themselves. Not infrastructure managing applications, but applications as autonomous agents that navigate decentralized infrastructure, find suitable nodes, and self-serve. Type-theory-based cloud primitives provide compile-time guarantees. DSP/FFT extracts runtime patterns.

I derive concepts from first principles, then validate against existing work. Read more about my methodology...

Current Work

Built with our libraries:

synapse

High-performance CUDA FFI library with optimized GPU kernels. Rust-safe bindings for matrix operations, memory management, and tensor core acceleration.

Rust + C++/CUDA · Production FFI patterns · MIT license

axon Exploratory

Routing layer for multi-cloud LLM inference. Currently a thin vLLM wrapper. The abstraction becomes valuable when Synkti expands beyond AWS to GCP, Azure, etc. Waiting for real usage patterns before building routing features.

InferenceBackend trait · Multi-cloud potential · BSL-1.1 license

Expertise

Rust Async Runtimes (tokio, async_stream) Distributed Systems State Machine Design CUDA/C++ GPU Programming

Engagements

Available for CUDA-Rust consulting on high-performance systems, GPU orchestration, and distributed infrastructure.

Links