I build local and hybrid AI systems that turn reusable techniques into agent workflows and long-horizon knowledge architectures.

My work sits at the intersection of engineering discipline, agent systems, and living architectures of thought.

I build workflows, infrastructure, and reusable techniques that make collaboration between humans and agents more legible, reproducible, and safe.

Two long-horizon directions organize most of what I build:

• Agents of Abyss (AoA): an evolving ecosystem of agents, workflows, memory, routing, and infrastructure
• Tree of Sophia (ToS): a living knowledge architecture for philosophy and world thought

For a compact ecosystem vocabulary, see GLOSSARY.md.

• for the ecosystem center, layer map, and federation rules, start with Agents-of-Abyss
• for the knowledge world and long-horizon architecture of thought, start with Tree-of-Sophia
• for reusable practice, execution workflows, and proof surfaces, move through aoa-techniques -> aoa-skills -> aoa-evals
• for the runtime body beneath AoA and ToS, go to abyss-stack

• growing Tree-of-Sophia as a long-horizon knowledge architecture for philosophy and world thought
• establishing Agents-of-Abyss as the ecosystem center and constitutional map of AoA
• publishing reusable techniques in aoa-techniques
• building aoa-skills as the execution layer for bounded agent workflows
• shaping aoa-evals as the proof layer for bounded agent quality and behavior
• preparing aoa-routing as the navigation and dispatch layer across AoA surfaces
• bootstrapping aoa-memo as the memory and recall layer for explicit, reviewable memory surfaces
• bootstrapping aoa-agents as the role and persona layer for explicit agent contracts
• bootstrapping aoa-playbooks as the scenario and composition layer for recurring operational recipes
• bootstrapping aoa-kag as the derived knowledge substrate layer for provenance-aware knowledge structures
• developing abyss-stack as the modular local and hybrid AI foundation beneath AoA and ToS

• agent systems that stay legible as they scale
• knowledge architectures that accumulate layers without collapsing into noise
• infrastructure where new tools become durable capabilities
• smaller and larger models using the same layered AoA surfaces effectively

• publish durable techniques, not one-off accidents
• build systems that stay legible, reviewable, and reproducible
• let new tools become new layers, not chaos multipliers
• prefer modular growth over brittle fusion
• meaning for humans, acceleration for agents

• humans and agents in the loop, with clear boundaries and handoffs
• reviewable workflows over opaque automation
• layered systems that can evolve without losing their source of truth

• Tree-of-Sophia
  Living knowledge architecture for philosophy and world thought

• Agents-of-Abyss
  Ecosystem center for AoA: charter, layer map, federation rules, and program-level direction

• aoa-techniques
  Reusable techniques for coding agents and humans

• aoa-skills
  Bounded agent-facing execution workflows built on top of reusable techniques

• aoa-evals
  Portable evaluation bundles for agents and agent-shaped workflows

• aoa-routing
  Emerging navigation and dispatch layer for routing models and humans across AoA surfaces

• aoa-memo
  Memory and recall layer for explicit, reviewable, provenance-aware memory surfaces

• aoa-agents
  Role and persona layer for explicit agent contracts, boundaries, and handoff posture

• aoa-playbooks
  Scenario and composition layer for recurring operational recipes across AoA surfaces

• aoa-kag
  Derived knowledge substrate for provenance-aware, graph-friendly knowledge structures

• abyss-stack
  Modular local and hybrid AI foundation for agent workflows, memory, data, and observability

• ATM10-Agent (private, in development)
  Local multimodal companion with perception, memory, safe automation, and voice

• Systems: Fedora, Podman, Windows 11
• Languages: Python, Bash, JavaScript, PowerShell
• Agent & app layer: FastAPI, Uvicorn, Streamlit, LangChain, LiteLLM, n8n
• Models & inference: OpenVINO, OpenVINO GenAI, Ollama, Transformers, Torch
• Data & memory: Postgres, Redis, Neo4j, Qdrant
• Observability: Grafana, Prometheus, Alertmanager
• Build workflow: ChatGPT, Codex, GitHub

• LinkedIn