Financial Services

NEFT (National Electronic Funds Transfer): Batch-based, settled in hourly batches, available 24x7 since 2019, any amount. RTGS (Real Time Gross Settlement): Real-time, individual transaction settlement, minimum ₹2 lakh, used for high-value payments. IMPS (Immediate Payment Service): Real-time like RTGS but for any amount (up to ₹5 lakh), 24x7, widely used for retail. UPI is built on IMPS rails. Key difference: RTGS is gross settlement (each transaction settled individually), NEFT is net settlement (batched). RTGS has higher finality guarantee.

Account Aggregator (AA) is RBI's consent-based financial data sharing framework. It allows customers to share their financial data (bank statements, insurance policies, investments) with Financial Information Users (FIUs — like loan providers) via Financial Information Providers (FIPs — like banks) through a secure, consent-managed flow. Key players: Finvu, OneMoney, PhonePe AA, CAMS Finserv. Use case: Instead of uploading bank statements for a loan, share 12 months of data digitally in seconds. Built on DEPA (Data Empowerment and Protection Architecture) principles.

A UPI-style payment system requires several key components: (1) API Gateway for authentication and routing of payment requests. (2) Payment Service — a microservice that validates balances, debits the sender via Core Banking System, and orchestrates the transfer. (3) Routing Layer (like NPCI's UPI Switch) that identifies the destination bank from the VPA handle and forwards the request. (4) Receiving Bank CBS that validates the recipient and credits the account. (5) Settlement Engine for reconciliation — real-time or batch depending on the rail. (6) Notification Service for push/SMS confirmations. Key design considerations: idempotency (prevent double payments), eventual consistency across distributed ledgers, timeout/retry handling for network failures, and audit logging for every transaction step.

Clearing is the process of aggregating, reconciling, and preparing payment instructions before the actual transfer of funds. It involves netting multiple transactions (e.g. Bank A owes Bank B ₹100, Bank B owes Bank A ₹60 — net: Bank A pays Bank B ₹40). Settlement is the final, irrevocable transfer of funds between institutions to complete the transaction. In NEFT, clearing happens in batches (every 30 min) and settlement follows. In RTGS, each high-value transaction is settled immediately (gross settlement). In UPI, settlement is effectively real-time — both debit and credit happen within seconds, with end-of-day reconciliation.

A Policy Administration System (PAS) is the central system of record for all insurance policies. It manages the complete policy lifecycle: (1) Quoting — calculating premiums based on risk parameters. (2) Underwriting — applying business rules to decide coverage terms. (3) Policy Issuance — creating the formal policy with coverage, beneficiaries, and premium schedule. (4) Mid-term changes — endorsements, coverage modifications. (5) Renewals — auto-renewal or lapsed policy handling. (6) Cancellations. The PAS integrates with Claim Management Systems (for adjudication), Billing Systems (for premium collection), and Regulatory Reporting systems (for IRDAI filings). Examples: Majesco, LifePRO, Guidewire PolicyCenter.

Financial systems use event-driven architecture (EDA) because transactions are naturally event-based — a payment debit, a trade execution, a policy renewal are all discrete events that multiple downstream systems need to react to. Benefits: (1) Loose coupling — the Payment Service doesn't need to know about the Fraud Service or Reporting Service; it just publishes an event. (2) Real-time processing — the Fraud Service can analyze a transaction event instantly. (3) Resilience — if the Reporting Service is down, it processes events when it recovers (using Kafka's durable log). (4) Scalability — each consumer scales independently. (5) Audit trail — event logs provide an immutable history. Common tools: Apache Kafka for high-throughput streaming, RabbitMQ/IBM MQ for reliable queuing.

KYC (Know Your Customer) is the process of verifying a customer's identity before allowing them to use financial services. Technically it involves: document collection (Aadhaar, PAN, passport), OCR/AI-based document verification, biometric matching, and integration with government identity registries (e.g. UIDAI for Aadhaar). The KYC status is stored in the CRM and checked before account activation. AML (Anti-Money Laundering) involves detecting suspicious transaction patterns. Technically: (1) Transaction monitoring rules engine (e.g. flagging cash deposits above thresholds). (2) Watchlist screening against sanctions lists (OFAC, UN). (3) Machine learning models for anomaly detection (unusual transaction frequency, amounts, geographies). (4) Suspicious Activity Reports (SARs) filed with FIU-India. Both are regulatory requirements under RBI/PMLA guidelines.