Building a consumer carpooling platform for Indian highways requires solving a difficult distributed systems challenge: how do you verify thousands of documents in seconds with banking-grade reliability while safeguarding user privacy?
This technical overview outlines the system architecture powering WayloShare's Two-Way KYC Engine, our asynchronous microservice pipeline, and how we leverage IndiaStack APIs for real-time document validation.
System overview & security goals
The WayloShare verification service was engineered around three strict technical tenets:
- Sub-90-second Verification Latency: Users expect onboarding to complete rapidly without waiting hours for manual agent approval.
- Zero Plaintext Storage: Government document numbers (UIDAI Aadhaar tokens, Driving Licence IDs) must never be stored unencrypted in application databases.
- Anti-Spoofing & Liveness Assurance: The engine must reject pre-recorded videos, static photo screenshots, and deepfake camera injections.
The 4-stage automated KYC pipeline
When a user initiates profile verification, the client SDK opens a secure TLS 1.3 session with our Verification Gateway. The request flows through an event-driven state machine:
│
▼
[Stage 1: Document OCR Ingestion] ──► Extracts Name, DOB, Gender, Reg No
│
▼
[Stage 2: IndiaStack Gateway] ──► Validates DigiLocker / Parivahan Registry
│
▼
[Stage 3: Biometric Liveness] ──► Passive Face Liveness & 3D Depth Check
│
▼
[Stage 4: Tokenization & Vault] ──► Issues Verified Badge Token; Masks PII
IndiaStack integration: DigiLocker & Parivahan
Rather than relying on unverified image uploads that can be tampered with in image editors, WayloShare interfaces directly with authorized Open APIs:
- DigiLocker OAuth Gateway: Fetches cryptographically signed XML documents directly from the issuing authority, eliminating document forgery.
- Parivahan Vahan & Sarathi Connectors: Real-time lookup of commercial vs private vehicle class, road-worthiness fitness certificate, and valid driving license categories (LMV / Transport).
Zero plaintext storage & tokenized isolation
To adhere to the Digital Personal Data Protection (DPDP) Act, user documents are segregated in hardware-isolated vaults. Sensitive document numbers are hashed using salted SHA-256 and stored as non-reversible verification tokens.
Once a profile is confirmed, client-facing microservices only receive a boolean verification flag (is_verified: true, kyc_tier: full). Personal identifiers are never returned in public API payloads, keeping user identities safe from scraping or unauthorized access.