dsf-enginexai-sdk 2.0.0

Explainable AI Scoring SDK with LLM Integration for Transparent Risk Assessment

DSF EngineXAI SDK

Configurable Evaluation Engine with AI Integration

Enterprise analysis module that combines structured data with AI-generated signals while maintaining transparent audit trails.

---

🚀 Why EngineXAI?

Balance flexibility and transparency in evaluation workflows:

• Configure logic without hardcoding
• Integrate AI-generated signals under client control
• Maintain mathematical transparency

---

🧠 AI Signal Integration

Incorporate AI-generated assessments as configurable inputs.

config = {
    "ai_assessment": {
        "input_type": "external",
        "params": {"tuning": 2.5}
    },
    "metric_primary": {
        "input_type": "standard",
        "reference_value": 700,
        "params": {"tuning": 2.0}
    }
}

applicant = {
    "ai_assessment": 0.73, 
    "metric_primary": 680
}

---

⚙️ Installation

pip install dsf-enginexai-sdk

---

🧩 Quick Start

from dsf_enginexai_sdk import CreditScoreClient

config = {
    "metric_a": {
        "input_type": "standard", 
        "reference_value": 3000, 
        "params": {"tuning": 1.8}
    },
    "metric_b": {
        "input_type": "standard", 
        "reference_value": 0.3, 
        "params": {"tuning": 2.5}
    }
}

subject = {"metric_a": 2800, "metric_b": 0.42}

with EvaluationClient(api_key="your_key", tier="community") as client:
    result = client.evaluate(subject, config)
    print(f"Outcome: {result['outcome']}")
    print(f"Score: {result['score']:.4f}")

---

🤖 AI Integration Pattern

ai_score = your_ai_system.assess(data)


subject = {
    "ai_signal": ai_score,
    "structured_metric": 720
}

result = client.evaluate(subject, config)

Note: AI signals are client-provided. EngineXAI does not generate or store AI content.

---

🔍 Analysis Breakdown

Available for Professional and Enterprise tiers.

result = client.evaluate(subject, config, audit=True)
breakdown = result["audit_trail"]

Example:

[
  {
    "feature": "ai_signal",
    "contribution_pct": 35.2,
    "metadata": "optional"
  },
  {
    "feature": "metric_primary", 
    "contribution_pct": 28.5
  }
]

---

📊 Tier Comparison

Feature	Community	Professional	Enterprise
AI Signal Input	✅	✅	✅
Batch Size	1	100	500
Audit Trails	❌	✅	✅
Adaptive Mode	❌	✅	✅

---

🧬 Advanced Features

Adaptive Evaluation (Pro/Enterprise)

result = client.evaluate_batch(
    subjects, 
    config,
    adaptive={'enabled': True}
)

Automatically adjusts for missing fields.

---

🎯 Use Cases

AI-Enhanced Analysis

config = {
    "ai_context": {"input_type": "external", "params": {...}},
    "traditional_metric": {"input_type": "standard", "reference_value": 1.0, "params": {...}}
}

Hybrid Systems

config = {
    "standard_metric": {"input_type": "standard", "reference_value": 700, "params": {...}},
    "ml_output": {"input_type": "external", "params": {...}},
    "ai_context": {"input_type": "external", "params": {...}}
}

---

⚠️ Important Notes

Client Responsibility:
Clients must validate compliance with applicable regulations. This SDK is an evaluation support tool.

Data Processing:
All logic executes server-side. SDK exposes only configuration interface.

Input Requirements:
Inputs must be normalized where applicable. See technical documentation (requires NDA).

AI Signals:
AI-generated signals are client-provided and client-controlled. EngineXAI does not generate, interpret, or store AI content.

---

📞 Support

Licensing: [e-mail]
Technical Documentation: Available under NDA

---

🔒 Architecture

• Server-side computation engine
• Configurable evaluation rules
• Transparent audit interface
• Secure API authentication

Technical specifications available under NDA.

---

📋 Credits

Technology Architect: Jaime Alexander Jimenez

---

© 2025 DSF UpTech. All rights reserved.