Serverless Bank Statement Extraction & OCR Engine
verified
Sub-Second OCR Table Extraction
An enterprise automated document extraction engine designed for high-precision parsing of financial statements and identity documentation. Built on AWS serverless primitives, it leverages AWS Textract for OCR and utilizes an asynchronous AWS Step Functions callback pattern to orchestrate multi-page PDF ingestion, validation, and human-in-the-loop review.
- Architecture: Fully serverless EventBridge and Step Functions orchestration handling fluctuating ingestion volume with zero idle compute cost.
- Extraction Precision: Customized table parsing algorithms that normalize erratic statement formats into structured, schema-validated JSON data.
- Infrastructure as Code: 100% defined using AWS CDK with TypeScript, ensuring identical staging and production environments.
AWS Textract
AWS Step Functions
AWS Lambda
AWS CDK
TypeScript
EventBridge
Adform Flow Unified Trading Graph
speed
Unified Enterprise API Federation
Unified GraphQL federation gateway consolidating disparate backend domain services across the global Adform Flow advertising ecosystem. Solved cross-squad schema friction by instituting strict schema governance, automated linting in CI/CD, and transparent developer experience platform services.
- Governance: Established automated schema composition checks preventing breaking changes from reaching production.
- Resilience: Containerized gateway deployed on Kubernetes with multi-layer caching on Redis to absorb high-frequency query bursts.
- Developer Velocity: Created centralized schema registries and mock tools, reducing integration time for client application teams by 35%.
Node.js
GraphQL Federation
Kubernetes
Apollo Gateway
Docker
MongoDB
Redis
DWP Searchlight Microservice Modernization
shield
15+ Core Services Containerized
Comprehensive modernization engagement targeting mission-critical civil service applications serving millions of UK citizens. Modernized 15+ monolithic or aging services into clean containerized microservices running on AWS EKS and serverless compute, backed by automated GitLab CI pipelines.
- Containerization: Standardized Docker container base images and secure runtime profiles meeting UK government security standards.
- Observability: Integrated centralized structured logging, tracing, and metric collection to identify latency bottlenecks under peak citizen usage.
- Zero-Downtime Delivery: Implemented rolling blue/green deployments, eliminating service interruption during system upgrades.
AWS EKS
AWS Lambda
Docker
GitLab CI
TypeScript
Node.js
CloudWatch
Multi-Platform Ecosystem: Android TV Display, iOS & Android Apps
tv
Synchronized Mobile & TV Fleet
A synchronized cross-platform community and facility management ecosystem engineered for West Midlands Community Centre (WMDC). Built as a monorepo architecture powering Android TV in-mosque timetable screens, community mobile applications across iOS and Android, and a serverless Firebase backend.
- TV Display App (Android TV / Large Screens): Real-time high-visibility digital signage with automated prayer countdowns, screen orientation controls, keep-awake runtime, and audio notifications.
- Community Mobile App (iOS & Android): Native mobile experience built with React Native & Expo featuring push notifications, background fetch, daily quotes, event announcements, donation gateways, and geolocation-based prayer calculators.
- Shared Monorepo Architecture: Turborepo modular architecture sharing battle-tested TypeScript packages across apps (
prayer-lib, event-lib, donate-lib, quote-lib).
- Serverless Cloud Backend: Automated Firebase Cloud Functions in Node.js orchestrating timetable data feeds, CSV parsing, and push notification dispatches.
React Native (0.81)
Expo 54
Android TV
iOS & Android
Firebase Functions
Turborepo
TypeScript
Shared Libraries