Update README.md

README.md

@@ -1,18 +1,29 @@
 # 🏭 ExpertData-Factory
-**Industrial-Scale High-Fidelity Reasoning Data**
+**Expert-verified reasoning datasets for LLM fine-tuning**
 
-ExpertData-Factory is a specialized data refinement lab dedicated to generating elite **Chain-of-Thought (CoT)** datasets for the next generation of LLMs. We focus on high-rarity niches where reasoning is the primary bottleneck for model performance.
+ExpertData-Factory is a data refinement lab building high-fidelity reasoning datasets for post-training, alignment, and evaluation. We specialize in **high-rarity domains** where structured reasoning quality is the primary bottleneck for model performance.
 
-## 🧪 Our Methodology
-Our "Alchemist" pipeline transforms raw technical documentation and scientific papers into structured reasoning assets using a multi-stage verification process:
-1. **Extraction**: Automated mining from expert-grade sources.
-2. **Refinement**: Transforming data into logical CoT structures.
-3. **Verification**: 100% Ground-Truth validation using state-of-the-art embedding models (`text-embedding-005`).
-4. **Sanitization**: Rigorous PII scanning and redaction for enterprise safety.
+## ✅ What you get
+- **Schema-stable** reasoning records (ideal for fine-tuning + eval)
+- **PII scanning & redaction** for enterprise safety
+- **Verification layer** using embedding-based grounding + consistency checks (`text-embedding-005`)
+- Exports: **JSONL** and **Parquet**
 
-## 🎯 Key Domains
-* **Cybersecurity**: Deep threat logic, vulnerability analysis, and MITRE-aligned reasoning.
-* **Scientific Reasoning (Upcoming)**: Methodological logic, hypothesis validation, and experimental analysis (Rarity Score 1.0).
+## 🧪 Methodology (Factory Pipeline)
+Our "Alchemist" pipeline transforms expert-grade sources (technical docs + scientific papers) into structured reasoning assets through a multi-stage QA process:
 
-## 💼 Enterprise & Licensing
-We offer both **Public** samples for the community and **Gated/Commercial** datasets for enterprise fine-tuning. For specialized data mining requests in high-rarity niches, please contact us via the Hugging Face portal.
+1. **Extraction** — automated mining from expert-grade sources  
+2. **Refinement** — normalization into consistent reasoning structures  
+3. **Verification** — embedding-grounded checks + schema validation (powered by `text-embedding-005`)  
+4. **Sanitization** — rigorous PII detection and redaction
+
+## 🎯 Domains
+- **Cybersecurity (Public)** — threat logic, vulnerability analysis, and MITRE-aligned reasoning  
+- **Scientific Reasoning (Gated, launching soon)** — methods, causality, hypothesis testing, and experimental reasoning  
+  *(AI / Bio / Physics — public sample released first, full dataset via access request)*
+
+## 🔐 Enterprise & Licensing
+We provide **public samples** for community research and **gated/commercial** datasets for enterprise fine-tuning.
+
+**Request access / custom builds**
+- If you’re an enterprise or lab and want full gated access or custom data generation, **request access on the gated dataset page** or message the organization via Hugging Face.