Anum Abdullah

Problem

Sunesis ran on legacy, outdated architecture with no scalable structure. The team operated across three disconnected products, tracked delivery manually offline, and had no integration between systems — creating friction, duplicated effort, and a ceiling on growth.

Approach

• Migrated the platform from its legacy architecture to AWS, resolving scalability and cost constraints
• Owned discovery, backlog grooming, and stakeholder sign-off across 5 AI modules
• Shipped a learner onboarding chatbot
• Shipped a feedback/review AI that tells learners where they stand and what assessment to take next
• Shipped an at-risk learner flagging module for auditors, identifying candidates for audit review
• Shipped an LLM-based personalized learning path generator, sequencing content by learner interest, ranking, and progress
• Ran the full sprint cycle: backlog alignment, grooming, estimation, demos, and delivery to production

Outcome

Problem

A branchless, CBUAE-regulated digital bank needed to onboard customers at scale using OCR-extracted identity documents, while guaranteeing data privacy, secure third-party API integration, and system stability under high concurrent load — a first-attempt, regulator-facing go-live with zero margin for compliance error.

Approach

• Owned the central admin portal managing customer onboarding status and sensitive identity data
• Enforced OCR data validation to prevent information leakage during transaction processing
• Implemented API retry logic to prevent silent failures in third-party routing
• Built load balancing into the architecture to hold stability under high onboarding volume
• Drove UAT resolution on Emirates ID extraction accuracy and correct bank-routing logic ahead of go-live
• Supported a no-fee, flexible-FX onboarding offer to drive early adoption

Outcome

Problem

A sports-industry client tracked global events — World Cups, marathons, governing bodies, sponsorship and broadcast deals — manually, compiling everything into PDFs and brochures for third parties. There was no structured system to query, map, or report on this data by geography, sport, or commercial relationship.

Approach

• Phase 1: built automated web scraping and extraction to replace manual PDF cataloging
• Mapped data hierarchically by geography, sport category, governing body, and commercial deal
• Phase 2: delivered a custom client dashboard with configurable event metrics and automated reporting per event and deal
• Delivered a minimal, scalable core architecture built for extensibility

Outcome

Problem

Beneple ran on an unstructured, poorly optimized .NET legacy framework with no scalability and a dated UI/UX that fell short of modern usability and accessibility standards.

Approach

• Re-architected the system, cutting AWS costing by ~30%
• Delivered a modern, accessibility-compliant UI/UX
• Reduced SLA turnaround and optimized core processes
• Phase 2: expanded scope to salary management, recruitment, onboarding/offboarding, training, and assessment modules
• Refactored the full codebase to a scalable structure based on earlier evaluation findings

Outcome

Problem

US retail and hospitality brands needed a multi-brand, multi-branch inventory and stock management system — with invoice processing still handled entirely by hand across brands.

Approach

• Built a multi-tenant structure: each brand supports multiple branches, each managing its own stock, inventory, orders, and dispatch
• Delivered OCR-based invoice extraction with brand-specific dynamic field mapping
• Designed confidence-score routing: extractions at or above 80% confidence sync automatically to third-party systems; lower-confidence extractions route to human moderation
• Delivered a modern, user-friendly UX to support fast client onboarding and adoption

Outcome

Problem

Mayfair's portfolio of five projects and products had no centralized governance — no standardized reporting, documentation, or delivery visibility, making it difficult for leadership to track risk, status, or SLA performance across the business.

Approach

• Established standardized documentation: project plans, status sheets, backlogs, high-level roadmaps, and technical/AI documentation
• Built a recurring SLA reporting cadence, including dedicated incident reports for critical production issues
• Introduced monthly client reporting and board-level presentations across all projects
• Standardized process flow mapping across every client engagement
• Embedded MVP/POC frameworks into the delivery lifecycle to validate ideas before full build-out

Outcome