Text Generation
Transformers
English
aml
kyc
cdd
compliance
governance
audit
human-in-the-loop
risk-management
financial-crime
decision-support
policy-aware-ai
review-chain
bpm-red-academy
Instructions to use MightHubHumAI/FinC2E with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use MightHubHumAI/FinC2E with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="MightHubHumAI/FinC2E")# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("MightHubHumAI/FinC2E", dtype="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps
- vLLM
How to use MightHubHumAI/FinC2E with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "MightHubHumAI/FinC2E" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "MightHubHumAI/FinC2E", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/MightHubHumAI/FinC2E
- SGLang
How to use MightHubHumAI/FinC2E with SGLang:
Install from pip and serve model
# Install SGLang from pip: pip install sglang # Start the SGLang server: python3 -m sglang.launch_server \ --model-path "MightHubHumAI/FinC2E" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "MightHubHumAI/FinC2E", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker images
docker run --gpus all \ --shm-size 32g \ -p 30000:30000 \ -v ~/.cache/huggingface:/root/.cache/huggingface \ --env "HF_TOKEN=<secret>" \ --ipc=host \ lmsysorg/sglang:latest \ python3 -m sglang.launch_server \ --model-path "MightHubHumAI/FinC2E" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "MightHubHumAI/FinC2E", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use MightHubHumAI/FinC2E with Docker Model Runner:
docker model run hf.co/MightHubHumAI/FinC2E
| license: other | |
| language: | |
| - en | |
| tags: | |
| - aml | |
| - kyc | |
| - cdd | |
| - compliance | |
| - governance | |
| - audit | |
| - human-in-the-loop | |
| - risk-management | |
| - financial-crime | |
| - decision-support | |
| - policy-aware-ai | |
| - review-chain | |
| - bpm-red-academy | |
| pipeline_tag: text-generation | |
| library_name: transformers | |
| # FinC2E — Financial Cognitive Compliance Engine | |
| **Governance-first AML/KYC case structuring and review-chain intelligence for human-accountable institutional workflows.** | |
| FinC2E is part of the **BPM RED Academy — HumAI MightHub** governance stack. | |
| It is designed for regulated and high-accountability environments where AI must remain: | |
| - advisory-only | |
| - human-reviewed | |
| - policy-aware | |
| - explainable | |
| - traceable | |
| - audit-ready | |
| FinC2E is not positioned as an autonomous compliance decision-maker. | |
| It is designed to support structured AML/KYC/CDD case analysis, human review-chain governance, escalation handling, audit-ready reasoning, and controlled institutional evaluation. | |
| --- | |
| ## Product Context | |
| FinC2E supports the broader **FinC2E Studio** workflow: | |
| | Layer | Purpose | | |
| |---|---| | |
| | Case Intake | Structured AML/KYC/CDD fields, jurisdictions, UBO, source of funds, PEP, sanctions, adverse media, and narrative context | | |
| | Policy-Aware Analysis | Risk level, score, reason, controls, red flags, assumptions, missing information, audit note, committee summary | | |
| | Human Review Chain | Reviewer role, decision, notes, escalation, override, final disposition | | |
| | Lifecycle & Worklist | Case state, priority, worklist bucket, next required action | | |
| | Management Dashboard | Executive visibility over review status, closure state, attention level, and workflow posture | | |
| | Audit Trail | Event IDs, timestamps, actors, policy mode, source integrity, and event payload | | |
| | Export Pack | Executive Case Report, Review Chain Bundle, Management Summary, Audit Package | | |
| --- | |
| ## Validated System Status | |
| FinC2E Studio has completed **Golden Test Set v2** validation. | |
| | Validation Area | Status | | |
| |---|---| | |
| | Low-risk AML/KYC scenario | PASSED | | |
| | Medium-risk uncertainty scenario | PASSED | | |
| | High-risk missing UBO / unclear source-of-funds scenario | PASSED | | |
| | Critical sanctions / PEP / adverse media scenario | PASSED | | |
| | Contradiction handling | PASSED | | |
| | Escalation workflow | PASSED | | |
| | Override workflow | PASSED | | |
| | Final disposition and closure | PASSED | | |
| | Post-closure guardrail | PASSED | | |
| | Audit trail and export package generation | PASSED | | |
| **Current system status:** Validated pilot-ready governance product. | |
| --- | |
| ## Intended Use | |
| FinC2E is intended for controlled institutional evaluation in: | |
| - AML / KYC / CDD case structuring | |
| - suspicious transaction review support | |
| - UBO and source-of-funds gap analysis | |
| - PEP, sanctions, and adverse media triage support | |
| - red-flag extraction | |
| - controls recommendation support | |
| - committee briefing preparation | |
| - audit-ready narrative generation | |
| - review-chain governance | |
| - controlled compliance workflow pilots | |
| --- | |
| ## Not Intended For | |
| FinC2E is not intended for: | |
| - autonomous compliance decisions | |
| - autonomous transaction approval | |
| - autonomous rejection, blocking, freezing, penalties, or reporting | |
| - legal or regulatory advice | |
| - public consumer use | |
| - replacing compliance, audit, legal, or risk professionals | |
| - unrestricted use with real personal or confidential institutional data | |
| All outputs require qualified human review. | |
| --- | |
| ## Governance Boundary | |
| FinC2E follows a strict governance boundary: | |
| | Boundary | Position | | |
| |---|---| | |
| | Autonomous enforcement | Not supported | | |
| | Human accountability | Required | | |
| | Final decision authority | Remains with institution and qualified human reviewers | | |
| | Regulatory/legal authority | Not claimed | | |
| | Public real-data use | Not recommended | | |
| | Private institutional deployment | Required for real records | | |
| --- | |
| ## Output Orientation | |
| FinC2E is designed to support structured outputs such as: | |
| ```json | |
| { | |
| "risk_level": "Low | Medium | High | Critical", | |
| "score": 0.0, | |
| "reason": "short structured explanation", | |
| "recommended_action": "APPROVE | REVIEW | ESCALATE | REJECT | BLOCK_SAR", | |
| "controls": ["control 1", "control 2"], | |
| "red_flags": ["flag 1", "flag 2"], | |
| "assumptions": ["assumption 1", "assumption 2"], | |
| "missing_information": ["item 1", "item 2"], | |
| "audit_note": "audit-ready narrative", | |
| "committee_summary": "brief committee-ready summary" | |
| } | |
| This schema is intended to support human review, committee briefing, audit preparation, workflow integration, and institutional traceability. | |
| System Integration | |
| FinC2E is designed to operate as part of the FinC2E Studio workflow and may be combined with: | |
| Hugging Face Spaces | |
| model / fallback inference configuration | |
| policy profiles | |
| human review workflow | |
| lifecycle/worklist logic | |
| audit trail | |
| export packages | |
| private institutional deployment patterns | |
| future AI Factory governance workloads | |
| Controlled Pilot Orientation | |
| Recommended pilot format: | |
| Controlled AML/KYC Governance Pilot | |
| A 4–6 week advisory-only pilot using synthetic or anonymized AML/KYC cases to evaluate: | |
| case structuring quality | |
| risk posture consistency | |
| review-chain governance | |
| escalation and override handling | |
| management visibility | |
| audit package usefulness | |
| Recommended scope: | |
| 25–100 synthetic or anonymized cases | |
| 2–4 policy profiles | |
| human reviewer workflow testing | |
| audit package review | |
| pilot evaluation summary | |
| ## Related Assets | |
| - **FinC2E Studio:** https://huggingface.co/spaces/bpmredacademy/HumAI_FinC2E_HQ | |
| - **Governance Gateway:** https://huggingface.co/spaces/MightHubHumAI/FinC2E-Governance | |
| - **HumAI MightHub Organization:** https://huggingface.co/MightHubHumAI | |
| - **BPM RED Academy Website:** https://www.bpm.ba | |
| - **FinC2E Website Page:** https://www.bpm.ba/FinC2E | |
| - **Founder Profile:** https://huggingface.co/bpmredacademy | |
| Institutional Disclaimer | |
| FinC2E is provided for controlled evaluation, research, pilot preparation, and institutional workflow design. | |
| It does not provide legal advice, regulatory advice, financial advice, or autonomous compliance decisions. | |
| All outputs must be reviewed by qualified personnel under applicable institutional, legal, regulatory, and governance frameworks. | |
| BPM RED Academy — HumAI MightHub | |
| Engineering legitimacy into AI systems. | |