Spaces:
Running
Running
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,10 +1,75 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
-
|
| 3 |
-
|
| 4 |
-
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
---
|
| 9 |
|
| 10 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
title: README
|
| 3 |
+
emoji: 💻
|
| 4 |
+
colorFrom: purple
|
| 5 |
+
colorTo: green
|
| 6 |
+
sdk: gradio
|
| 7 |
+
pinned: false
|
| 8 |
+
sdk_version: 6.2.0
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
<div align="center">
|
| 12 |
+
|
| 13 |
+
# 🛡️ Cata Risk Lab
|
| 14 |
+
### Sovereign AI Assurance & Technical Gatekeeping
|
| 15 |
+
**Zurich • London • Miami**
|
| 16 |
+
|
| 17 |
+
[](https://opensource.org/licenses/MIT)
|
| 18 |
+
[](https://doi.org/10.5281/zenodo.xxxxxxx)
|
| 19 |
+
[](https://www.fedlex.admin.ch/eli/cc/2022/491/en)
|
| 20 |
+
[](https://artificialintelligenceact.eu/)
|
| 21 |
+
|
| 22 |
+
</div>
|
| 23 |
+
|
| 24 |
---
|
| 25 |
+
|
| 26 |
+
## 🏛️ The Mission: Zero-Trust AI
|
| 27 |
+
**Cata Risk Lab** is a specialized technical auditing firm focused on **Algorithmic Accountability** for high-stakes industries. We do not build "creative" AI; we build **Sovereign Gatekeepers** that prevent liability.
|
| 28 |
+
|
| 29 |
+
In an era of generative hallucination, our mandate is simple: **If the AI cannot cite the source, the AI must fail.**
|
| 30 |
+
|
| 31 |
+
### 🔬 Core Methodology
|
| 32 |
+
We operate on a **"Compliance-in-the-Loop"** architecture designed for:
|
| 33 |
+
* **Swiss Banking Standards (nFADP)**
|
| 34 |
+
* **EU AI Act (High-Risk Classification)**
|
| 35 |
+
* **UK & Australian Privacy Principles**
|
| 36 |
+
|
| 37 |
---
|
| 38 |
|
| 39 |
+
## 🚀 Deployed Architectures
|
| 40 |
+
|
| 41 |
+
### [🛡️ Veritas: The Hallucination Auditor](https://huggingface.co/spaces/Cata-Risk-Lab/Veritas-Auditor)
|
| 42 |
+
*Status: Live (v1.0.0)*
|
| 43 |
+
|
| 44 |
+
Veritas is our flagship **RAG (Retrieval Augmented Generation) Auditor**. It ingests financial and regulatory documentation (PDFs) and answers queries with strict adherence to the source text.
|
| 45 |
+
|
| 46 |
+
* **Zero-Trust Retrieval:** Returns explicit "FAIL" verdicts for missing data.
|
| 47 |
+
* **Audit Trail:** Generates citation-backed logs.
|
| 48 |
+
* **Tech Stack:** `LangChain` | `Azure OpenAI` | `ChromaDB` | `Python`
|
| 49 |
+
|
| 50 |
+
> *"A guess in a regulatory audit is not a glitch. It is a lawsuit."*
|
| 51 |
+
|
| 52 |
+
---
|
| 53 |
+
|
| 54 |
+
## 🛠️ Technology Stack
|
| 55 |
+
We build using "Industrial-Grade" open source and enterprise foundations:
|
| 56 |
+
|
| 57 |
+
| Layer | Technology |
|
| 58 |
+
| :--- | :--- |
|
| 59 |
+
| **Orchestration** | LangChain, LlamaIndex |
|
| 60 |
+
| **Inference** | Azure OpenAI (GPT-4o), Google Gemini 1.5 Pro |
|
| 61 |
+
| **Vector Search** | ChromaDB, Azure AI Search |
|
| 62 |
+
| **Deployment** | Microsoft Foundry, Hugging Face Spaces |
|
| 63 |
+
|
| 64 |
+
---
|
| 65 |
+
|
| 66 |
+
## 📬 Contact & Governance
|
| 67 |
+
**Cata Risk Lab** operates as a technical due diligence partner for Enterprise Clients.
|
| 68 |
+
|
| 69 |
+
* **GitHub:** [github.com/dcata004](https://github.com/dcata004)
|
| 70 |
+
* **Verification:** DOI Minted via Zenodo for all Audit Artifacts.
|
| 71 |
+
|
| 72 |
+
---
|
| 73 |
+
<div align="center">
|
| 74 |
+
<i>"Sovereignty is the new Compliance."</i>
|
| 75 |
+
</div>
|