Independent AI engineer building distributed autonomous systems on multi-GPU Linux clusters. I design architectures where hundreds of AI agents coordinate, trade, speak, and self-improve — without human intervention.
Core domains: Multi-agent orchestration | Voice interfaces | Algorithmic trading | Browser automation | GPU cluster engineering
A 9-layer autonomous operating system spanning boot to voice, running 600+ AI agents across a 6-GPU cluster.
| Metric | Value |
|---|---|
| 🤖 Autonomous Agents | 600+ MCP handlers across distributed nodes |
| 🎙️ Voice Commands | 2,658 recognized commands, Whisper CUDA pipeline (<300ms) |
| 🔧 MCP Tools | 87 tools orchestrated via Claude Agent SDK |
| 🔗 Domino Chains | 835 automation pipelines, self-healing |
| 📈 Trading Engine | 6-model consensus, MEXC Futures, 800+ pairs |
| 🧠 Inference | 6 NVIDIA GPUs / 46GB VRAM, LM Studio + Ollama |
| Project | Description | Stack |
|---|---|---|
| jarvis-linux | 🧠 Core OS — cognitive cluster, orchestration, voice, trading | Python, TS |
| jarvis-cowork | 🏭 570+ autonomous QA scripts, continuous self-repair | Python |
| jarvis-whisper-flow | 🎙️ Real-time Whisper CUDA voice pipeline (<300ms latency) | Python, CUDA |
| TradeOracle | 📈 Multi-model consensus trading engine — MEXC Futures | Python |
| TradeOracle-Nexus-Elastic | 🔍 Financial intelligence — Elasticsearch, Monte Carlo sims | Python |
| browser-mcp-orchestrator | 🌐 Dual-browser DevTools MCP orchestration | Node.js |
| bibliotheque-prompts-multi-ia | 📚 397+ optimized prompts for Claude, GPT, Gemini, Mistral | Markdown |
| turbo | 📊 Cluster dashboard — GPU monitoring, agent health | Python |
┌─────────────────────────────────────────────────────────┐
│ M1 │ Ryzen 5700X3D │ 6 GPUs │ 46GB VRAM │ Primary: inference, voice, orchestration
│ M2 │ 3 GPUs │ 24GB VRAM │ Reasoning: DeepSeek-R1
│ M3 │ Remote node │ Backup │ Failover reasoning
│ OL1 │ Ollama local │ Lightweight │ Fast tasks, cloud fallback
└─────────────────────────────────────────────────────────┘