Upay Mobile Financial Services

Upay is a prominent Mobile Financial Service (MFS) in Bangladesh by UCB Fintech Company Limited. It processes massive volumes of transactions, requiring absolute precision, strict compliance, and ultra-low latency.

As a Senior Backend Engineer & Solution Architect, I designed and architected the high-throughput payment infrastructure and universal financial aggregator.

---

🏛️ Major Architectural Pillars

As a Senior Backend Engineer & Solution Architect, I designed the payment infrastructure and led the delivery of Upay’s core transaction microservices:

1. High-Throughput Transaction Orchestration Pillar

• Subsystems Included: Basic MFS Transaction Gateway Real-Time Scope & Limit Engine
• Tech Stack: Golang Echo PostgreSQL (ULID) Redis Distributed Systems
• Architecture & Leadership Impact:
  • Scatter-Gather Concurrency: Parallelized non-dependent downstream API calls (sender/receiver profiles, limit checks, fee calculations) using Goroutines and channels to eliminate network I/O blocking.
  • Saga Pattern Microservices Orchestration: Implemented strict compensation logic. If any downstream service or local DB commit fails after limits are reserved, asynchronous reversal routines trigger automatically to guarantee distributed transaction consistency without 2PC locks.
  • PostgreSQL ULID Migration: Replaced UUIDs/auto-increment IDs with ULIDs (Universally Unique Lexicographically Sortable Identifiers) to eliminate B-Tree index fragmentation under peak write loads.

2. Universal Financial Aggregation & DPS Pillar

• Subsystems Included: Universal Financial Aggregator (DPS Service) Partner Bank Middleware
• Tech Stack: Python/Django Strategy Pattern Adapter Pattern PostgreSQL
• Architecture & Leadership Impact:
  • Interface-Driven Design: Designed a normalized financial aggregator using Strategy and Adapter patterns, reducing new partner bank onboarding time from months to days.
  • Lifecycle Management: Orchestrated complete DPS workflows—account creation, installment collections, encashments, and automated reconciliation.

3. Dynamic Rule Evaluation & Rate Engine Pillar

• Subsystems Included: Dynamic Charge & Commission Engine Utility Integration Gateway (BREB & Teletalk)
• Tech Stack: Python/Django Redis REST/SOAP Fintech
• Architecture & Leadership Impact:
  • Deterministic O(1) Caching: Built a runtime configuration layer for complex fee slabs (fixed amounts, percentages, max-caps, variable slabs), bypassing relational DB lookups via multi-tiered Redis caching.
  • Utility Protocol Translation: Led the implementation of custom protocol adapters for major utility providers (BREB, Teletalk), handling high-volume bill payment processing and automated reconciliation.