📍 St. George, UT ↔ Dallas, TX | 🤖 Agentic systems builder | 🚀 Founder @ Ruska AI
Deep in agent orchestration mode — building production-grade AI systems that don’t just chat, but run, coordinate, and ship.
Focused on turning LLMs into durable infrastructure instead of demos.
Agent-native software infrastructure for real businesses.
Ruska is an opinionated platform for:
- Multi-agent orchestration (LangGraph, DeepAgents, MCP, A2A)
- Persistent state & filesystem-backed context
- Sub-agents you can spin up, tear down, and route across tools
- One system → many surfaces: Web, IDE, CLI, TUI, Slack, CI
🌐 https://ruska.ai
📺 Building in public on YouTube, Twitch, and LinkedIn.
- 🎼 Orchestra – Multi-agent orchestration engine with checkpoints, queues, and tool routing
- 🧑
✈️ DeepAgents Harness – Long-running agents that can delegate, pause, and resume work - 🖥️ Agent Chat UI – Self-hosted, auth-ready agent interface (LangGraph-native)
- 🧰 Ruska CLI – Manage agents, prompts, tasks, and state from your terminal
- 🗂️ Persistent Context Layer – File-system + DB backed memory for agents that outlive sessions
- 🔌 MCP Servers & Bridges – Claude Code, Codex, tools, and agents talking to each other
- Agent Frameworks: LangGraph, LangChain, DeepAgents, MCP
- Backend: FastAPI, TaskIQ, Redis, Postgres, Supabase
- Infra: Docker, AWS, Bedrock, Cloudflare, Netlify
- Models: Claude, GPT-4.x, Gemini, local & hosted LLMs
- Dev Ergonomics: CLIs, TUIs, streaming APIs, checkpointing, resumability
I care a lot about:
- Deterministic agent behavior
- Observability over “vibes”
- Systems that survive restarts, crashes, and humans
- 🎥 Prompt Engineers AI – Nightly stream building agents in public
- 👥 Plano Prompt Engineers – 1,000+ member AI builder community
- ✍️ Writing about:
- Agent orchestration
- Production LLM systems
- Building leverage as a solo / small-team engineer
- 🧠 Teaching devs how to stop prompting and start architecting
- Shipping Ruska as a unified agent control plane
- Designing IaC for agents (spin up, edit, destroy, rehydrate)
- Hardening long-running agents with queues + checkpoints
- Exploring agent → agent → agent workflows at scale
- Helping teams replace brittle SaaS workflows with AI-native systems
“Always look for two confirmations.”
- Agents should be auditable, not magical
- Shipping > theorizing
- AI is leverage, not a replacement for taste
- Build systems that work at 2am when nobody’s watching
Random Facts
- Treat agents like distributed systems, not chatbots
- Prefer CLIs over dashboards (until dashboards earn it)
- Believe “agent infra” will be as normal as web servers
- Build in public to stay honest
- Clarity > doomscrolling (most days)