AI Engineering, Production Stack & Tools
The tools and methods I actually build with — agent orchestration, verifiable contracts, spatial 3D interfaces and sub-second voice pipelines. It is written down so that, if you want to, you can see exactly what your project would be built on.
Agent Systems & Orchestration
Multi-agent coding swarms over isolated git worktrees, intelligent role-to-model routing, and unified tool exposure via Model Context Protocol.
Evaluation & Benchmark Design
Anti-hallucination evaluation protocols, Pareto frontier ranking across trade-offs, and fail-closed validity gating on every autonomous run.
Real-Time Low-Latency Voice AI
Bi-directional WebSocket streaming connecting telecom audio to Deepgram STT, streaming LLM inference, and Cartesia TTS with conversational barge-in.
Answer Engine Optimization (AEO)
Technical AEO workflow treating AI-crawler policies, llms.txt exports, and structured JSON-LD graphs as one unified, citable system.
The AI Stack in Production
The exact engineering tiers, runtimes, and protocols deployed across our client systems and videos:
Flagship Open Source & Tooling
Public frameworks, skills, and tools backing our technical claims.
Gravity Skills
Evidence-first, cost-aware engineering skills for Antigravity and compatible AI coding assistants — an open-source plugin for agentic workflows that stay practical, verifiable and proportionate to the task.
Antigravity Add Model
Proxy patch for Google Antigravity that intercepts its internal Cloud Code API and adds support for OpenAI, Claude, Ollama, Gemini and any custom LLM provider.
MRMP Agent MCP Server
Standard Model Context Protocol server allowing Claude and Cursor to autonomously query skills and code.
Need a high-performance AI system, 3D interface, or voice pipeline?
Hands-on 2 to 4 week execution sprints with verified deliverables. Architecture to production deployment.