> booting hypervisor... OK > syncing model weights... done (43.2GB) > initializing node cluster [ws-23]... online > deploying agentic runtime... complete > user: tom-boyle

Building distributed intelligence at the edge.
No frameworks, no boilerplate — just systems that run fast.

⚡ Core Stack

• Languages: TypeScript • Rust • Python • CUDA
• Infra: WSL2 • Docker • Pulumi • Vast.ai • Vercel • Warp
• ML: PyTorch • JAX • ONNX • RAG • Agentic runtimes
• Editor: Neovim — no mouse.

---

🧱 Current Focus

• 🔥 Building agentic compute infrastructure — GPUs spun up in seconds, globally distributed, optimized for inference-heavy workloads.
• 🧠 Architecting runtime layers for LLMs that plan, act, and adapt.
• 🛰️ Experimenting with edge intelligence: low-latency model execution near sensors.

---

📂 Projects

🛰️ Swift Compute

AI GPU infrastructure for developers who don’t wait.

• Sub-minute H100 provisioning
• Transparent per-minute billing
• Distributed orchestration with Pulumi

🧠 Deep Runtime

Custom agentic runtime — zero frameworks, built for real-time decision systems.

• Structured memory
• Tool use
• Event-driven orchestration

---

🌐 Connect

• 🕸️ tomboyle.io
• 🐦 x.com/tomlikestocode
• 💼 linkedin.com/in/tom-boyle-au