Architecture and Components

🏗️ KYC Architecture and Components

📊 KYC Architecture Diagram

🏛️ KYC Architecture and Components

The KYC (Know Your Customer) system is a modular platform for automated user identity verification using uploaded document images, selfies, and other sources. The system’s architecture flexibly adapts to various business scenarios and requirements: from banks and fintech to marketplaces and government services.

Supported Scenarios:

• Document — the user sends a photo of their passport, ID card, or other identification document.
• Selfie with document — verification of the user’s face against the photo in the document.
• Selfie — verification with a single facial image.
• Liveness check — confirmation that a live person is in front of the camera, not a photo or a fake.
• Combined scenarios — for example, document + selfie with document + Liveness.

Purpose of system components:

Each KYC component is responsible for a specific stage of processing and verification:

• Document Image Processing: detection, alignment, classification, and segmentation of the document into fields (photo, full name, dates, etc.).
• Optical Character Recognition (OCR): data extraction and checksum verification.
• Face image analysis: comparison of the face with the photo in the document, age and gender estimation.
• Anti-fraud mechanisms: detection of photocopies, screenshots, Photoshop, screens, physical wear.
• Liveness Check: determining if a live person is in front of the camera.
• Integrations with external systems: verification of the document and user against databases of the Ministry of Internal Affairs, FSSP, and blacklists.
• Decision Engine: the final stage where all results are aggregated and a decision is formed: “Approved”, “Rejected”, or “Requires further review”.

Thus, the KYC architecture is built around a set of independent, scalable, and interconnected modules, allowing it to be quickly adapted to any verification requirements and scenarios.

🗂️ Module Architecture and Interaction

User sends images (document + selfie with document) via WebSDK/API
                                  |
                                  v
                           +---------------+
                           |  API Gateway  |
                           +-------+-------+
                                   |
   +-----------------+-------------+-------------+------------------+
   |                 |             |             |                  |
   v                 v             v             v                  v
+---------------+ +---------------+ +------------------+ +------------------+
|  Document     | |  Selfie       | |  Liveness Check  | | External         |
|  Detector     | |  Detector     | | (liveness check) | | Databases        |
|(document photo)| |(selfie w/doc.)| |                  | |(MVD, FSSP, etc.) |
+-------+-------+ +-------+-------+ +---------+--------+ +---------+--------+
        |                 |                   |                    |
        v                 v                   |                    |
+-----------------+ +-------------------+     |                    |
| Quality Check   | | Selfie Check      |     |                    |
|(format, size,   | | - Face Matching   |     |                    |
| illumination,   | | - Gender, age     |     |                    |
| alignment)      | | - Document Detector|     |                    |
+-------+-------+ |                   |     |                    |
| --- |
        v           |                   |     |                    |
+-----------------+ +---------+---------+     |                    |
|Document Type    |           |               |                    |
|Classification   |           |               |                    |
+-------+---------+           |               |                    |
| --- |
        +-----------+         |               |                    |
        |           |         |               |                    |
        v           v         v               v                    v
+--------------+ +------------------+ +-----------------+ +----------------+
| Document     | | Anti-fraud checks| | Liveness Check  | | External DB    |
| Segmenter    | |(screens, copy,  | | Result          | | Check Result   |
|(fields, photo,| |editor, wear and | |                 | |                |
| signature,   | |tear, logic)      | +--------+--------+ +-------+--------+
|stamp)        | +---------+---------+          |                  |
+------+-------+                    |                  |
| --- |
       v                   v                    |                  |
+--------------+    +-----------------+         |                  |
|      OCR     |    | Anti-fraud      |         |                  |
|(recognition, |   | Check Results   |         |                  |
|sticker/       |   |                 |         |                  |
|overprint check)|   +-----------------+---------+------------------+
+------+--------+                    |
       |                             |
       +-------------+---------------+
                     |
                     v
              +----------------+
              | Decision Engine|
              |(Decision Center)|
              +--------+-------+
                       |
                       v
                 +-------------+
                 |  Result     |
                 |(Approved /  |
                 | Rejected /  |
                 |Additional   |
                 |check)       |
                 +------+------+
                        |
                        v
                 +---------------+
                 |  API Gateway  |
                 +-------+-------+
                         |
                         v
                   User

1. Document Detection

Responsible for checking the presence of a document in the image, photo quality, and its preparation for further processing.

Functions:

• File format check (JPEG, PNG)
• Check image size and resolution
• Check for lighting and absence of shading
• Document alignment and normalization on the image

2. Selfie with Document Detection

Responsible for verifying selfies with documents, matching user data, and preventing fraud.

Functions:

• Comparison of the face on the document and the selfie (Face Matching)
• Verification of user’s age and gender against document data
• Check for signs of screens (photos from a phone screen, monitor)
• Check for signs of graphic editor tampering
• Check for signs of photocopies or printouts
• Document detection on selfie and its type determination

3. User Liveness Check

Features:

• Analysis of serial images or video for signs of a living person
• Discarding photos or records from screens and paper media
• Deepfake detection in photos and videos The neural network identifies signs of generative editing, replaced faces, “face swapping,” and GAN artifacts
• Detection of fully generated images Identification of photos created by neural networks (GAN/Stable Diffusion/FaceApp, etc.), including inconsistencies in textures, lighting, depth, and generation artifacts

4. External Database Verification

Functions:

• Document and Identity Status Check
• Check passport validity against official sources
• Driver’s license check
• Comprehensive passport check (series/number, loss, blocks)
• Debt and Financial Checks
• Check for debts of individuals in the FSSP
• Check tax debts
• Check the taxpayer status of NPD (self-employed)
• Financial scoring of credit bureaus
• Short-term loan default scoring
• Collateral check
• Check the activity and expiration date of the phone number
• Full name–phone binding confirmation
• Judicial and legal checks
• Check against the list of bankrupts
• Search for bankruptcy cases in arbitration courts
• Check arbitration cases
• Checking cases in courts of general jurisdiction
• Check against the database of nominal directors
• Check communication with legal entities
• Wanted check: — Ministry of Internal Affairs — Federal Bailiff Service — Federal Penitentiary Service
• Identity Verification Checks
• SNILS ↔ Full Name Compliance Check
• Obtaining TIN by full name and passport data
• Check email and full name compliance
• Check the use of the number by a specific individual
• Sanctions and Restrictions
• Check against the list of terrorists
• Sanctions screening of individuals (EU, UK, OFAC, UN, etc.)
• Check blocked accounts for sole proprietors
• Check in the registry of foreign agents
• Check in the registry of alimony debtors

5. Document Classification & Segmentation

Determines the type of the uploaded document and highlights key elements for further processing.

Functions:

• Document classification (passport, driver’s license, ID card, etc.)
• Segmentation of the document into separate areas (text fields, photograph, signatures, stamps)

6. Document Field Recognition and Verification (OCR & Field Verification)

Recognizes document text and verifies the authenticity of its individual elements.

Functions:

• OCR recognition of text from segmented fields
• Check for inserts or drawings in the document fields
• Checksum verification (e.g., MRZ code)

7. Document Antifraud Checks

Responsible for identifying fraud attempts and document forgery.

Functions:

• Checking photos of documents taken from device screens (mobile, monitors, TVs)
• Identification of signs of photocopies or printed documents
• Check for graphic editor interference (field editing)
• Check the degree of document deterioration (damage, signs of wear)
• Check for logical inconsistencies in document fields (dates, numbers, series)

8. Decision Engine

Aggregates the results of all modules, makes the final decision on the KYC procedure.

Functions:

• Collection and analysis of verification results (selfie with document, OCR, anti-fraud, Liveness, external databases)
• Issuance of the final decision based on the results of the inspection: “Approved”, “Rejected”, or “Requires additional inspection”
• Generation of error codes and descriptions (if any)