🧠 Sigma-Zero: Industrial Quality AI Orchestrator

An advanced, deterministic Multi-Agent Orchestration system engineered to autonomously execute the Six Sigma DMAIC (Define, Measure, Analyze, Improve, Control) methodology over manufacturing defect data, fusing industrial quality rigor with modern AI vector retrieval.

🏗️ System Architecture & Workflow

Sigma-Zero replaces manual process analysis and generic LLM guessing with a deterministic data validation pipeline and an embedded Graph-like Vector Memory (sqlite-vec), ensuring 100% data-anchored, zero-hallucination root cause synthesis.

graph TD;
    subgraph Data Ingestion [Production Data Ingestion]
        A[Manufacturing Logs CSV] --> B{Pydantic Immutable Validator};
        B -->|Clean DAG| C(Pareto Analysis Engine);
    end

    subgraph Cognitive Core [Six Sigma Analytics]
        C -->|Defect Topology| D[Process Capability Engine Cpk/Cp];
        D -->|Sigma Level| E[sqlite-vec Vectorized Hindsight Memory];
    end

    subgraph Sovereign Memory [Historical Knowledge]
        E <--> F[(sqlite-vec Deep Memory)];
    end

    subgraph Edge Compute [Report Synthesis]
        E -->|Context Injection| G[Ockham Synthesizer & Reporter];
        G --> H[Markdown & HTML Output];
    end

Loading

🚀 Technical Highlights

• Deterministic Ingestion & Validation: Uses Pydantic models as the "Iron Truth" to validate production data types and constraints before processing, eliminating silent downstream failures.
• Process Capability Engine (Cpk): Automated statistical engine (using pandas and numpy) that mathematically calculates $Cp$, $Cpk$, and the $Sigma\ Level$ of the production line.
• Embedded Vector Knowledge Base (sqlite-vec): A fully local, serverless vector database embedded directly in SQLite. It queries historical defects and known solutions instantly without network overhead or heavy vector databases (Pinecone, Milvus).
• Zero-Hallucination Synthesis: Employs an adversarial data-grounding approach where the LLM (or deterministic templates) acts strictly as a formatter over mathematical truths and historically retrieved vectors.
• Portable CLI Orchestration: Managed via uv (ultrafast python package manager) to guarantee 100% reproducibility and hermetic environments across any Linux edge machine.

🛠️ Stack & Dependencies

• Core Engine: Python 3.12, pydantic (Deterministic schemas), pandas
• Analytics & Vis: numpy, scipy, plotly
• Edge AI & Memory: sqlite-vec (Embedded C++ WASM vector search)
• Environment: uv, dotenv

⚙️ Setup & Execution

1. Requirements

Ensure uv is installed on your system.

curl -LsSf https://astral.sh/uv/install.sh | sh

2. Execution

Run the zero-touch automated demo script. This will populate the simulated SQLite-vec knowledge base, calculate capabilities, and generate the final report.

cd sigma-zero
chmod +x demo.sh
./demo.sh

3. Testing

Execute the pytest suite to validate the statistical and validation engines:

uv run pytest tests/

---

Built for rigorous Personal Knowledge Graph compilation and deterministic quality engineering.