I design the workflows — AI implements them.
Available for contracts, consulting, and full-time roles.
Async-first · Local-first · Deterministic-by-design
🌐 Landing Page · 𝕏 Twitter · 🦙 Ollama · 📧 Email
I'm a 36-year-old single father and primary caregiver for my young son with autism. I walked away from my career 4 years ago because he needed someone present — and spent those 4 years teaching myself AI from the ground up, starting with hardware, because my long-term goal is total resource self-sufficiency with AI.
Background: Precision machinist (±0.00015" tolerances) → Contract IT → Independent AI Systems Architect
Education: B.A. Psychology — George Mason University
Philosophy: I treat AI model failures as diagnostic problems, not tuning problems. My psychology training taught me behavioral diagnostics — isolating variables, identifying failure patterns, tracing root causes. I apply that same methodology to agent systems.
I build governed orchestration systems that turn probabilistic AI models into reliable, observable, operator-controlled workforces. My work focuses on:
- AI Agent Systems Architecture — coordination layers, tool governance, observability pipelines
- Local-First AI Infrastructure — Ollama clusters, model routing, fallback chains, distributed inference
- Agent Safety & Observability — zero-silent-action policies, audit trails, real-time monitoring
I specialize in systems that other engineers find too complex to own.
|
⭐ Sheppard AI agent for Ollama handling memory, automation, and knowledge distillation using Redis, PostgreSQL, and ChromaDB. |
SOLLOL Performance-aware load balancing for distributed Ollama clusters with Ray and Dask. |
|
BrokeLLM Local control plane for model-slot routing, fallbacks, and provider switching across CLI coding tools. |
AgentFabric Distributed CLI orchestration framework for multi-agent systems with governed MCP architecture. |
|
Amnesic Distributed memory architecture for agent coordination with externalized long-term reasoning state. |
Vigilance Real-time observability and audit system enforcing zero-silent-action policy for AI agents. |
|
LlamaForge LoRA fine-tuning pipeline with CPU/GPU DDP, GGUF conversion, and Ollama integration. |
Conflux AI development accelerator for structured multi-agent coding workflows with deterministic task graphs. |
| Category | Technologies |
|---|---|
| Languages | Python, JavaScript/TypeScript |
| AI/ML | Ollama, PyTorch, DDP, LoRA, GGUF, RAG |
| Infrastructure | Ray, Dask, Docker, Redis, PostgreSQL, ChromaDB |
| Architecture | MCP, Agent Orchestration, Distributed Systems |
I'm open to contract work, consulting engagements, and full-time roles in:
- AI infrastructure and orchestration systems
- Agent workflow design and governance
- Local-first AI deployment and optimization
- Developer tooling and observability
📧 [email protected]
🌐 bamnlanding.lovable.app
"I spent four years machining parts at ±0.00015" tolerances. I bring that same precision, constraint awareness, and failure-resilient design to every system I architect."