PROPELOO

FINTECH / FINANCIAL INFRASTRUCTURE

Build financial infrastructure designed for real-time money movement.

PROPELOO engineers FinTech systems — from payment processing and core ledgers to KYC/AML pipelines, fraud detection and the compliance infrastructure that regulators actually inspect. Moving money is 30 lines of code. Building the trust infrastructure around it is a separate engineering problem.

Moving money is easy. Building the trust infrastructure around identity, compliance and reconciliation isn't.

A payment gateway is 30 lines of code. The risk engine behind it is 30,000. The audit trail that satisfies a regulator is another product entirely. The reconciliation system proving every transaction balanced — built last, tested least — is where FinTech products fail in production. PROPELOO designs the full financial infrastructure stack: ledger, compliance layer, fraud detection and reconciliation. Not just the payment flow.

What sits underneath a production FinTech system.

A FinTech product is not a payment form. It is a ledger, a compliance system, a fraud engine and a reconciliation pipeline — each with its own failure modes and regulatory surface.

System Layers

  • Identity / Compliance Layer: KYC document verification, liveness checks, AML screening, sanctions watchlist, accreditation, ongoing monitoring
  • Transaction Processing Layer: Payment initiation, routing, authorisation, retry logic, idempotency, settlement, webhook delivery
  • Core Ledger Layer: Double-entry accounting, balance calculation, transaction history, immutable audit trail, multi-currency
  • Risk & Fraud Layer: Velocity rules, device fingerprinting, ML fraud scoring, manual review queue, chargeback management
  • Reporting & Audit Layer: Regulatory reporting, AML transaction monitoring, suspicious activity reports, data residency, compliance dashboard

Core Technical Capabilities

  • Payment Processing Infrastructure

    Multi-rail payment acceptance and disbursement — card networks, ACH/SEPA, wire transfer, crypto rails — with idempotent transaction processing, retry logic, webhook delivery and settlement reconciliation.

  • KYC/AML & Identity

    End-to-end identity verification pipeline — document checking, liveness detection, accreditation verification, sanctions screening (OFAC, UN, EU), PEP screening and ongoing transaction monitoring.

  • Transaction Ledger & Reconciliation

    Double-entry accounting ledger with immutable transaction records, real-time balance calculation, automated reconciliation against payment provider settlement files and discrepancy alerting.

  • Fraud Detection System

    Layered fraud detection — rule-based velocity checks for known patterns, ML model scoring for novel fraud, device fingerprinting, geolocation analysis and human review queues for high-risk transactions.

  • Banking-as-a-Service Integration

    BaaS provider integration (Unit, Synapse, Railsbank, Stripe Treasury) — account creation, card issuing, ACH origination, ledger synchronisation and regulatory passthrough compliance.

  • Regulatory Compliance Layer

    Compliance infrastructure for operating jurisdictions — AML transaction monitoring, SAR filing workflows, regulatory reporting pipelines, PCI DSS scoping and audit-ready documentation.

How we think about FinTech infrastructure.

Moving money is straightforward. Building the trust infrastructure around identity, compliance, fraud and reconciliation isn't.

  • Compliance is architecture

    KYC, AML, PCI DSS and PSD2 requirements are not features you bolt on. They define the data model, the identity layer and the audit trail from day one. Retrofitting compliance onto a built system is one of the most expensive FinTech mistakes.

    Axiom:

  • The ledger is the source of truth

    A FinTech product is only as trustworthy as its ledger. Double-entry accounting, idempotent transactions and real-time reconciliation must be correct before scale makes errors expensive — and before a regulator asks why balances don't add up.

    Axiom:

  • Fraud is a day-one problem

    Fraud patterns emerge within weeks of launch. Detection models, velocity rules, device fingerprinting and manual review queues must be live before the first real transaction — not after the first chargeback spike.

    Axiom:

  • Reliability is a trust metric

    A payment that fails or a balance that displays incorrectly erodes user trust faster than any other product failure. Five-nines availability, graceful degradation on third-party outages and sub-3-second transaction confirmation are product requirements.

    Axiom:

The decisions that define a FinTech system.

These choices determine your regulatory surface, operational cost model and what your product can do in year two. Most cannot be changed without a full rebuild.

  • In-house ledger vs Banking-as-a-Service

    Impact: BaaS platforms reduce time-to-market significantly but create long-term dependency. In-house ledgers give full control but require accounting expertise and regulatory engagement.

    • In-house double-entry ledger — full control, highest build cost, full regulatory responsibility
    • BaaS (Unit, Synapse, Railsbank) — faster launch, abstracted compliance, vendor dependency and per-transaction cost
    • Stripe Treasury — simplest, US-only, deeply tied to Stripe ecosystem
    • Hybrid — BaaS for banking infrastructure, in-house ledger for product-level accounting
  • Real-time vs batch settlement

    Impact: Settlement timing determines liquidity requirements and user experience. Real-time settlement changes the fraud model — funds are immediately accessible and irreversible.

    • Real-time settlement (Stripe instant, RTP, FedNow) — instant finality, higher cost per transaction
    • Batch settlement (ACH, SEPA standard) — lower cost, T+1 or T+2 delay
    • Hybrid — real-time for consumer UX, batch for B2B and large disbursements
  • Single-currency vs multi-currency architecture

    Impact: Multi-currency requires FX hedging, multi-currency ledger design, exchange rate management and additional KYC requirements for cross-border transactions.

    • Single currency — simplest ledger, limited market, no FX risk management required
    • Multi-currency with real-time conversion — broader market, FX risk, complex accounting
    • Multi-currency with settlement in one currency — best of both, requires FX hedging strategy
  • Synchronous vs asynchronous payment processing

    Impact: Async payments require idempotency keys, robust reconciliation and client-side state management for pending/failed/completed states. Errors are harder to detect and trace.

    • Synchronous — blocking confirmation, simple client state model, latency cost
    • Async with webhooks — non-blocking, higher throughput, complex reconciliation and error handling
    • Event-sourced — full immutable audit trail, replay capability, highest complexity
  • Self-hosted KYC vs managed provider

    Impact: Managed KYC providers handle regulatory updates automatically. In-house KYC requires tracking regulation changes across all operating jurisdictions.

    • Managed KYC provider (Synaps, Jumio, Onfido) — fast integration, per-verification cost, data with provider
    • In-house KYC — full data control, high build cost, requires ongoing regulatory maintenance
    • Hybrid — provider for document verification, in-house identity registry
  • Card network vs ACH vs crypto payment rails

    Impact: Rail selection defines fee economics, settlement speed, chargeback exposure and the regulatory surface. Multi-rail increases acceptance but multiplies reconciliation complexity.

    • Card network (Visa/Mastercard) — universal acceptance, 1.5-3.5% fees, chargeback exposure
    • ACH/SEPA bank transfer — lower cost, slower settlement, no chargeback (push payment)
    • Crypto rails (USDC/stablecoin) — near-instant, low cost, limited merchant acceptance, regulatory complexity
    • Multi-rail — accept all, route by cost and speed, most complex to reconcile

What FinTech infrastructure becomes.

  • Payment Gateway / PSP

    Multi-rail payment processing with merchant dashboard, webhook delivery, reconciliation reporting, fraud scoring and PCI DSS-compliant card data handling.

  • Neobank / Digital Wallet

    Consumer account infrastructure — KYC onboarding, IBAN issuance via BaaS, card issuing, P2P transfers, FX, savings products and real-time balance on a compliant core banking layer.

  • P2P Lending Platform

    Credit origination, automated underwriting, loan servicing, repayment scheduling, collections logic and regulatory reporting for consumer or SME lending products.

  • Investment & Wealth App

    Investment account infrastructure — KYC/accreditation, custody integration, portfolio rebalancing, order routing, performance reporting and regulatory documentation.

  • Cross-border Payment Platform

    FX conversion, international wire routing, compliance with sending and receiving jurisdiction AML regulations, real-time FX rate management and payout rail optimisation.

  • Embedded Finance Product

    Financial product embedded in a non-financial platform — BaaS-powered accounts, cards and payment flows with white-label UX, multi-tenant compliance and revenue sharing infrastructure.

The FinTech infrastructure stack.

Technology selection follows regulatory posture, transaction volume and the payment rails required — not framework preference.

  • Backend & APIs

    Stack: Node.js, Go, Python, PostgreSQL, Redis, Kafka / SQS

  • Payment Providers

    Stack: Stripe, Adyen, Stripe Treasury, Unit (BaaS), Railsbank, Marqeta (Card Issuing)

  • KYC / AML

    Stack: Synaps, Jumio, Onfido, Chainalysis, Elliptic, Sardine

  • Fraud & Risk

    Stack: Sardine, Sift Science, Custom ML Models, MaxMind GeoIP, Device Fingerprinting, DataDome

  • Infrastructure

    Stack: AWS, Docker, Kubernetes, Terraform, Datadog, PagerDuty

  • Compliance & Reporting

    Stack: Custom AML Engine, Transaction Monitoring, Compliance Reporting API, Audit Trail DB, GDPR Tooling

Financial infrastructure security requires multiple independent layers.

A FinTech product handles real money. Security failures are not support tickets — they are financial losses and regulatory events.

  • PCI DSS Compliance

    Any system that touches card data requires PCI DSS scoping. Architecture decisions — tokenisation at point of entry, network segmentation, cardholder data environment boundaries — determine the PCI scope and the cost of the annual QSA assessment. Scope reduction is an architecture goal.

  • Transaction Fraud & AML

    Card fraud, account takeover, synthetic identity fraud and money laundering are active threats from day one of launch. Velocity rules, device fingerprinting, ML transaction scoring and real-time AML screening must be designed and tested before the first real user transacts.

  • Identity Theft & Synthetic Fraud

    Synthetic identity fraud — combining real and fabricated personal information — bypasses document-only KYC. Liveness checks, database cross-referencing, behavioural analysis during onboarding and ongoing transaction pattern monitoring are required countermeasures.

  • API Security for Payments

    Payment APIs are high-value attack targets. Every endpoint requires authentication, authorisation checks on the specific account being accessed, idempotency key validation, rate limiting and anomaly detection on request patterns. OWASP API Top 10 is the baseline.

  • Encryption at Rest and Transit

    Financial PII requires encryption at rest (field-level for sensitive fields), TLS 1.3 minimum for all transit, key management infrastructure that survives personnel changes, and data residency controls compliant with GDPR and applicable financial regulations.

  • Regulatory Audit Readiness

    Regulators expect immutable audit trails, complete transaction history, AML monitoring records and evidence of KYC procedures. These must be designed into the data model from day one — retrofitting audit trails onto a running financial system is the most expensive FinTech architecture mistake.

From financial product concept to compliant production system.

  1. 01. Compliance & Regulatory Scoping

    Jurisdiction analysis, regulatory surface mapping, licence requirements, KYC/AML provider selection and PCI scope definition before architecture is designed.

  2. 02. Architecture & Data Model

    System architecture, ledger design, transaction state machine, identity data model, fraud scoring architecture and API contract specification.

  3. 03. Payment Flow Engineering

    Core payment processing implementation — transaction initiation, routing, processing, settlement, reconciliation, idempotency and webhook delivery.

  4. 04. KYC/AML & Fraud Layer

    Identity verification pipeline, AML transaction monitoring, fraud scoring integration, manual review queue and ongoing compliance monitoring infrastructure.

  5. 05. Compliance Reporting

    Regulatory reporting pipelines, SAR workflow, audit trail verification, compliance dashboard and jurisdiction-specific report generation.

  6. 06. Security Audit

    PCI scoping verification, penetration testing of payment flows, API security review, fraud model validation and compliance documentation review.

  7. 07. Launch & Monitoring

    Production deployment, real-time fraud monitoring activation, reconciliation alerting, incident response runbook, on-call protocol and regulatory go-live checklist.

FinTech platforms we have shipped to production.

Three regulated FinTech products handling real financial transactions.

  • Ledger and accounts module replacing legacy core banking for a digital bank

    Challenge: Digital bank needed a modern double-entry ledger and accounts module replacing their legacy core — supporting multi-currency accounts, real-time balance updates, and ISO 20022 payment messages without a 2-year core banking migration.

    Architecture: Event-sourced ledger: every transaction is an immutable event. Current balance derived from event stream. Kafka for real-time transaction propagation. Redis cache for hot balances. ISO 20022 message adapter layer.

    Outcome: 2.1M accounts migrated, real-time balance latency <50ms, zero reconciliation breaks in 12 months post-launch, audit by central bank regulator passed.

  • Automated loan origination and underwriting platform for a digital lender

    Challenge: Digital lender needed an automated underwriting platform pulling open banking data, running credit models, generating loan offers, and completing the full origination workflow — ID verification to e-signature — in under 5 minutes.

    Architecture: Plaid for bank statement data, Python ML model for credit scoring (gradient boosted, 48 features), rule engine for regulatory compliance checks, DocuSign for e-signature, loan ledger for drawdowns and repayments.

    Outcome: Average origination time 4m 12s, 23% default rate reduction vs manual underwriting, £40M in loans originated in first year.

  • Multi-rail payment orchestration layer for a payment service provider

    Challenge: PSP needed a payment orchestration layer routing transactions across 6 rails (card, bank transfer, direct debit, crypto, BNP, wallet) with intelligent retry, fallback routing, and unified settlement reporting.

    Architecture: Rule-based routing engine: payment type + amount + country determines optimal rail. Retry queue with exponential backoff. Fallback rail on failure. Unified settlement ledger normalising across rail-specific formats.

    Outcome: 94% first-attempt success rate (up from 81%), £120M monthly transaction volume, settlement reconciliation reduced from 8 hours to 12 minutes.

Frequently Asked Questions

Do we need a financial licence to launch?

It depends on the product and jurisdiction. Payment facilitation, money transmission and banking products have specific licence requirements in most jurisdictions. Many early-stage FinTech products launch under a BaaS provider's licence (Unit, Railsbank, Stripe Treasury) or as a non-custodial payment facilitator that does not require a money transmitter licence. We map the regulatory requirements for your specific product and geography in the first engagement week and recommend the appropriate structure.

What is the difference between a PSP, a bank and a payment gateway?

A payment gateway handles transaction transmission and authorisation — the technical layer between the merchant and the acquirer. A PSP (payment service provider) combines gateway, merchant account and settlement in one product. A bank (or BaaS-backed neobank) holds deposits, issues accounts and can originate ACH and wire transfers. The architecture you need depends on whether you are facilitating payments for third parties, holding user funds, or providing accounts. These have different regulatory requirements.

How do you implement KYC and AML?

KYC implementation: document upload + liveness check via a managed provider (Synaps, Onfido, Jumio), results stored in your identity database, onboarding gated on KYC pass. AML implementation: transaction monitoring rules (velocity, amount thresholds, geographic patterns) plus continuous sanctions screening against OFAC, UN and EU watchlists. Suspicious activity triggers a human review queue and potentially a SAR filing. The exact requirements depend on jurisdiction and product type.

Do we need to build a core banking ledger?

Most early-stage FinTech products should not build a custom core banking ledger. BaaS providers (Unit, Synapse) provide ledgering as part of their offering. The right time to build in-house is when BaaS constraints are limiting product features, the per-transaction cost at scale is prohibitive, or specific accounting requirements (multi-currency, custom waterfall logic) cannot be expressed in the BaaS model. We design a migration path from BaaS to in-house ledger when the business case is proven.

How do we support multiple currencies?

Multi-currency requires: a ledger designed with currency as a first-class entity on every balance and transaction record, FX conversion logic with rate sourcing and spread management, settlement in each currency or conversion strategy, multi-currency reconciliation and tax reporting considerations per jurisdiction. The architectural decision is whether to settle in a single base currency (simpler accounting, FX risk at conversion) or maintain balances in each currency (more complex, no conversion risk until payout).

How do you detect fraud at early stage when you have no data?

Early-stage fraud detection relies on rule-based systems while you accumulate the transaction data needed for ML models. Start with: velocity rules (5+ transactions in 10 minutes), device fingerprinting, IP geolocation checks, card BIN validation and known bad actor list screening. Integrate a managed fraud provider (Sardine, Sift) for their industry-wide signals. As your transaction volume grows past 50,000/month, your own ML fraud model becomes viable. The architecture must support this evolution without a rebuild.

What does PCI DSS compliance require?

PCI DSS compliance scope depends on how you handle card data. If you tokenise at point of entry using a managed provider (Stripe, Adyen) and never store raw card data, your PCI scope is SAQ A — the lightest tier. If you process raw card data, you face SAQ D or a full QSA assessment. Architecture decisions — tokenisation strategy, network segmentation, cardholder data environment boundary — determine PCI scope. Scope minimisation is an engineering goal, not a compliance afterthought.

How do you approach open banking and API integration?

Open banking integrations (Plaid, TrueLayer, Tink, direct PSD2 APIs) provide read access to account data and initiation of payments from user bank accounts. Implementation requires: OAuth2 connection flow for user authorisation, webhook handling for account updates, normalisation layer across different bank API formats, token refresh management and error handling for the high rate of open banking API failures. We design open banking integrations as first-class infrastructure, not a third-party API call.

How does reconciliation work for a FinTech product?

Reconciliation is the process of proving that every transaction in your ledger matches a real movement of money confirmed by the payment provider. It runs at settlement: compare your internal transaction records against the provider's settlement file, flag mismatches, investigate discrepancies and resolve any unmatched transactions. At early stage this can run daily. As volume grows, real-time reconciliation with automated discrepancy alerting is required. A FinTech product without automated reconciliation will have undetected errors.

How long does a FinTech product take to build?

A production payment product with KYC, ledger, fraud detection and compliance reporting takes 16-28 weeks from architecture to go-live. BaaS-backed products launch faster (10-16 weeks) because the banking infrastructure is abstracted. Products requiring custom card issuing or money transmission licensing take longer due to the compliance review cycles. We deliver in milestone-based sprints with staging environment demos at each stage.