ununseptium 1.0.0

State-of-the-art RegTech and Cybersecurity Python library for KYC/AML automation, data security, and AI-driven risk analysis

Ununseptium

State-of-the-Art RegTech and Cybersecurity Python Library

---

Table of Contents

• Overview
• Installation
• Quick Start
• Architecture
• Security Posture
• Auditability
• Mathematical Foundations
• Model Zoo
• CLI Reference
• Documentation
• Contributing
• License
• Citation

---

Overview

Ununseptium is an enterprise-grade Python library providing comprehensive tooling for:

Domain	Capabilities
KYC Automation	Identity verification, document processing, sanctions screening, entity resolution
AML Monitoring	Transaction analysis, typology detection, case management, regulatory reporting
Data Security	PII detection/masking, encryption, access control, tamper-evident audit logs
AI Risk Engine	Feature engineering, ensemble models, explainability, model governance
Scientific ML	Physics-Informed Neural Networks, Neural ODEs, Neural Operators
Math/Stats	Conformal prediction, EVT, Hawkes processes, copulas, graph statistics

Scope

Ununseptium provides computational primitives and frameworks for building compliance systems. It is designed for:

• Financial institutions implementing AML/KYC programs
• RegTech vendors building compliance platforms
• Research teams developing risk quantification methods
• Auditors requiring tamper-evident data trails

Non-Goals

• Ununseptium is NOT a turnkey compliance solution
• It does NOT provide legal advice or regulatory interpretation
• It does NOT replace human judgment in compliance decisions
• It is NOT a database or case management system (integrate with your own)

---

Installation

Requirements

• Python 3.11 or higher
• NumPy, SciPy, Pydantic v2

From PyPI

pip install ununseptium

From Source

git clone https://github.com/olaflaitinen/ununseptium.git
cd ununseptium
pip install -e ".[dev]"

Optional Dependencies

# Cryptography extras
pip install ununseptium[crypto]

# Full ML stack
pip install ununseptium[ml]

# All optional dependencies
pip install ununseptium[all]

---

Quick Start

Identity Verification

from ununseptium.kyc import Identity, IdentityVerifier

identity = Identity(
    full_name="John Doe",
    date_of_birth="1985-03-15",
    nationality="US",
    document_number="AB123456",
)

verifier = IdentityVerifier()
result = verifier.verify(identity)

print(f"Status: {result.status}")
print(f"Risk Level: {result.risk_level}")
print(f"Reason Codes: {result.reason_codes}")

Transaction Monitoring

from ununseptium.aml import Transaction, TypologyDetector

transactions = [
    Transaction(sender="ACC001", receiver="ACC002", amount=9500),
    Transaction(sender="ACC001", receiver="ACC003", amount=9600),
    Transaction(sender="ACC001", receiver="ACC004", amount=9400),
]

detector = TypologyDetector()
alerts = detector.detect(transactions)

for alert in alerts:
    print(f"Typology: {alert.typology_type}, Score: {alert.score}")

Tamper-Evident Audit

from ununseptium.security import AuditLog

log = AuditLog()

log.append({"action": "identity_verified", "identity_id": "ID-001"})
log.append({"action": "risk_assessed", "score": 0.72})

# Verify integrity
assert log.verify() is True

# Save with hash chain
log.save("audit.log")

---

Architecture

graph TB
    subgraph "Public API"
        CLI[CLI Interface]
        SDK[Python SDK]
    end

    subgraph "Domain Modules"
        KYC[KYC Module]
        AML[AML Module]
        SEC[Security Module]
    end

    subgraph "Intelligence Layer"
        AI[AI Module]
        MATH[MathStats Module]
        ZOO[Model Zoo]
    end

    subgraph "Infrastructure"
        CORE[Core Module]
        PLUG[Plugin System]
    end

    CLI --> KYC
    CLI --> AML
    SDK --> KYC
    SDK --> AML
    SDK --> SEC

    KYC --> AI
    AML --> AI
    AML --> MATH

    AI --> ZOO
    AI --> CORE
    MATH --> CORE

    SEC --> CORE
    PLUG --> CORE

Module Summary

Module	Purpose	Key Components
core	Foundation	Configuration, errors, logging, schemas
kyc	Identity	Verification, documents, screening, entity resolution
aml	Transactions	Monitoring, typologies, cases, reporting
security	Protection	PII, crypto, access control, audit logs
mathstats	Statistics	Conformal, EVT, Hawkes, copulas, graphs
ai	Intelligence	Features, models, explainability, governance
model_zoo	Pretrained	Catalog, download, verification
cli	Interface	Commands for all operations
plugins	Extensibility	Plugin discovery and loading

See docs/architecture/overview.md for detailed architecture documentation.

---

Security Posture

Ununseptium implements defense-in-depth security:

Data Protection

Layer	Mechanism	Implementation
Detection	PII Scanner	Regex patterns with configurable rules
Masking	Tokenization	Replace PII with reversible tokens
Encryption	Fernet/AES	Symmetric encryption with key rotation
Access	RBAC	Role-based permission enforcement

Integrity Assurance

The audit subsystem uses cryptographic hash chains:

$$H_n = \text{SHA256}(H_{n-1} | \text{entry}_n)$$

Where each entry is linked to its predecessor, making tampering detectable.

Threat Model

Threat	Mitigation
PII Exposure	Detection + masking pipeline
Unauthorized Access	RBAC with audit logging
Data Tampering	Hash-chain verification
Model Manipulation	Checksums + provenance tracking

See docs/security/security-overview.md for the complete threat model.

---

Auditability

Every operation in ununseptium can be traced through the audit subsystem:

from ununseptium.security import AuditLog, AuditVerifier

# Create verifiable audit trail
log = AuditLog()
log.append({"action": "screening_completed", "matches": 0})

# Later: verify integrity
verifier = AuditVerifier()
result = verifier.verify_file("audit.log")

if not result.is_valid:
    raise SecurityError(f"Tamper detected at entry {result.failed_index}")

Audit Entry Schema

Field	Type	Description
id	string	Unique entry identifier
timestamp	ISO-8601	Entry creation time
action	string	Action performed
actor	string	Who performed the action
resource	string	Affected resource
prev_hash	string	Hash of previous entry
entry_hash	string	Hash of this entry

---

Mathematical Foundations

Ununseptium provides statistically rigorous methods:

Conformal Prediction

Coverage-guaranteed prediction sets:

$$P(Y \in C(X)) \geq 1 - \alpha$$

from ununseptium.mathstats import ConformalPredictor

predictor = ConformalPredictor(alpha=0.1)
predictor.calibrate(y_cal, y_pred_cal)

pred_set = predictor.predict(y_new)
print(f"Interval: [{pred_set.lower}, {pred_set.upper}]")

Extreme Value Theory

Tail risk via Generalized Pareto Distribution:

$$F(x) = 1 - \left(1 + \xi \frac{x}{\sigma}\right)^{-1/\xi}$$

Sequential Detection

Real-time change detection with CUSUM, SPRT, and ADWIN algorithms.

See docs/mathstats/mathstats-overview.md for complete mathematical documentation.

---

Model Zoo

Pretrained models for common AML/KYC tasks:

Model ID	Domain	Architecture	AUC-ROC
aml-transaction-risk-v1	AML	Gradient Boosting	0.92
anomaly-detector-v1	Anomaly	Ensemble	0.88
entity-resolution-v1	KYC	Neural Network	0.91
sar-priority-v1	AML	Transformer	0.87
graph-risk-v1	AML	GNN	0.90

from ununseptium.model_zoo import PretrainedModel

model = PretrainedModel.load("aml-transaction-risk-v1")
result = model.predict(features)

See docs/model-zoo/model-zoo.md for the complete model catalog.

---

CLI Reference

# Display library information
ununseptium info

# Run diagnostics
ununseptium doctor

# Verify identity from JSON
ununseptium verify identity data.json --output result.json

# Screen a name
ununseptium screen name "John Doe" --threshold 0.8

# Verify audit log integrity
ununseptium audit verify audit.log

# Show audit entries
ununseptium audit show audit.log --limit 20

# Validate model card
ununseptium model validate model_card.json

---

Documentation

Comprehensive documentation is available in the docs/ directory:

• Architecture Overview
• KYC Module
• AML Module
• Security Guide
• Mathematical Methods
• AI and ML
• Model Zoo
• FAQ
• Glossary

---

Contributing

We welcome contributions. Please read CONTRIBUTING.md for:

• Development setup
• Code style guidelines
• Testing requirements
• Pull request process

---

License

Ununseptium is licensed under the Apache License 2.0. See LICENSE for details.

---

Citation

If you use ununseptium in academic work, please cite:

@software{ununseptium2025,
  author = {Laitinen, Olaf},
  title = {Ununseptium: RegTech and Cybersecurity Library},
  year = {2025},
  url = {<https://github.com/olaflaitinen/ununseptium>},
  version = {1.0.0}
}

See CITATION.md for additional citation formats.

---

Support

• GitHub Issues for bug reports
• GitHub Discussions for questions
• See SUPPORT.md for additional resources

---

Disclaimer: Ununseptium provides computational tools only. It does not constitute legal, compliance, or regulatory advice. Users are responsible for ensuring their implementations meet applicable regulatory requirements. Consult qualified legal and compliance professionals for guidance.