Project for UK based startup
Automating Fintech Lending Operations
TL;DR
A multi-stream transformation to streamline loan origination and underwriting for a fintech startup. We standardized and automated key workflows (income and expense validations, borrower bank account ownership, and debt consolidation), integrated core systems (Morpheus, Trinity, Experian, Neo), strengthened UAT and unhappy-path coverage, and explored targeted GenAI use. Result: faster assessments, fewer manual steps, and a scalable foundation for new products and markets.
The Challenge
- Multiple, fragmented workflows for loan processing across regions created inefficiencies and quality gaps.
- Manual, Excel-heavy validation of household income and expenses increased error risk and review cycles.
- Inconsistent borrower bank account verification and debt consolidation processes across markets.
- Dispersed governance and UAT practices led to slower feedback loops and unresolved unhappy-path scenarios.
The Solution
- Target architecture and platforms
- Core systems: Mambu (case and document flow), Broker Portal, Pricing Engine
- Integrations: Experian (credit bureau and Delphi scoring), Open Banking for bank data access, Onfido for ID&V documents
- Data and validation
- Automated checks for household income and expenses, mandatory data enforcement
- Borrower bank account ownership validation approach for joint vs single accounts
- Credit and policy rules including LTV, portfolio limits, CBTL exclusions, valuation policy by AVM or RICS
- Underwriting and decisioning
- Unhappy-path handling for failed credit search, max LTV breaches, ineligible categories and guardrails
- GenAI exploration
- Implementation of GenAI-assisted Q&A against underwriting guides with scoring on relevance, accuracy, clarity and action tracking for guide updates
The Process
Delivering Complexity Through Structure
Team setup
- Cross-functional squad including Business Analysts, backend and frontend engineers, QA, UAT coordinators, and product stakeholders;
- Regional stakeholders engaged from Ireland, London, and CH for requirement alignment
Cadence and ceremonies
- Sprint-based delivery; regular refinement sessions and end-of-sprint demos
- Weekly alignment meetings and show-and-tell, with risks and dependencies tracked; governance through analysis reviews and roadmap follow-ups
- UAT standups and defect triage
Internal processes
- Backlog management for client changes, policy document updates, and broker feedback incorporation
- Environment coordination with partner teams, and release management checkpoints
The Results
-
Significant gains in operational efficiency: Automated validations and reengineered workflows dramatically accelerated loan processing times, enabling faster decision-making and reducing processing bottlenecks across departments.
-
Enhanced data integrity and reduced manual workload: By minimizing human input in financial assessments, we achieved higher data accuracy and consistency, while freeing up teams to focus on high-value tasks instead of repetitive checks.
-
Superior user experience and adoption: Streamlined processes and seamless integration of tooling led to greater clarity, fewer user errors, and improved satisfaction among both internal users and external stakeholders.
-
A scalable, future-ready platform: The new architecture lays a robust foundation for continued innovation — supporting expansion into GenAI-driven decision support, deeper automation, and readiness for entry into new markets and product verticals.