Eurisko Mobility
Dec 2021 – Present
Engineer across web (React, TypeScript), mobile (React Native), and backend (Java Spring, Node NestJS, Python FastAPI). I lead delivery on teams of 2 to 5 engineers, work directly with business stakeholders, product owners, and tech leads, and own UI/UX direction. I also built the team's AI coding standards: one shared set of 7 rules, 14 skills, and 17 guides wired into Cursor, Claude Code, and GitHub Copilot, so every developer gets the same guidance whichever tool they use. That grew into ai-kit, a tool that sets up a new project with the skills, rules, and folder structure its stack needs. A feature that once took the whole team now ships end to end by one developer, with review focused on architecture, helped by a verification layer of TDD, visual testing and coding standards.
React, TypeScript, React Native, Java, Spring Boot, Node.js, NestJS, CI/CD, GitLab, Jenkins
Riyad Bank / JEEL (KSA)
5 projectsWallet-as-a-Service SDK
Floor 1 of 5React Native, iOS, Android, Flutter, Web
users
developers
team setup, was ~8 sprints
- Built a shared SDK for one of Saudi Arabia's largest banks, packaging common bank flows (transactions, cards, authentication and authorization, contact and location access) across iOS, Android, Flutter and web for 3 products serving 3.5M+ users, built by 15+ developers
- Took per-team setup, integration and repeated requirements explanation from roughly 8 sprints to 2-3 days by shipping a configuration guide
- Built a library-agnostic layer separating app components from the third-party libraries underneath, turning a dependency swap from a multi-sprint refactor into days
Relationship Manager Workbench & Mortgage Platform
Floor 2 of 5React, Redux, Java Spring
branches
less repetitive UI code
faster loads, at best
CI pipeline
- Developed internal platforms standardized across all 226 Riyad Bank branches, serving relationship managers and mortgage teams on a 115,000-line codebase
- Implemented a fully configurable dynamic table system full-stack using the factory pattern, removing ~75% of repetitive UI code
- Improved load times up to 90% and took CI pipeline duration from 45 to 15 minutes through test parallelization and targeted caching
Kids E-Banking
Floor 3 of 5React Native
fewer mobile API calls
- Owned biometrics, authentication and code optimization on a kids e-banking app, cutting mobile API calls by ~50% with client-side caching and request de-duplication
Online Mortgage
Floor 4 of 5ReactJS
- Led requirements planning, set review standards, documented features and infrastructure, and supported onboarding of new team members
API Management Platform
Floor 5 of 5ReactJS
- Defined frontend architecture and implementation approach; enforced testing and CI quality gates to support parallel delivery
Global consulting firm, ministerial client (KSA)
1 projectAI-driven Decision Support Platform
Ground floorReactJS, FastAPI, Python, Redis, MongoDB
job failure rate
frontend modules
- Designed a queue-based processing model to handle concurrent AI workloads, reducing failure rates from about 50 percent to around 10 percent through retries and database-backed state management
- Implemented event-driven communication between backend and frontend to surface real-time processing status in the UI
- Owned frontend development for more than 10 modules covering audio and video workflows, prompt editing, and feedback loops
Global consulting firm (MENA)
3 projectsAI Executive-Support Platform
Floor 1 of 3React, TypeScript, FastAPI, CrewAI, Redis, Qdrant
planning prep, was weeks
fewer model calls per message
less stored documents
API fields kept
- Led development of a multi-agent AI platform for a government executive office, cutting ministerial planning preparation from weeks to around 30 minutes
- Streamed per-stage progress and plain-language failures in place of static loaders, after users said they could not tell what was running
- Improved AI agent chat orchestration, cutting model calls by 50% per message by replacing a per-reply planner with a scoped context agent
Investment Feasibility Platform
Floor 2 of 3React, Python, Azure
accuracy gate in CI
- Delivered an AI-augmented platform for the firm's own consulting teams, automating the research and financial modeling behind feasibility studies on projects worth millions to billions
- Gated accuracy to 0.5% with CI checks that fail on regressions, holding parity between the platform's calculations and the consultants' Excel models
- Built human-in-the-loop correction so each study's adjustments feed into the next
Economic Estimation Automation
Floor 3 of 3Python, Oracle Cloud Kubernetes
- Automated gross-value-added estimation pipelines in Python, replacing manual, spreadsheet-driven estimation work with a repeatable pipeline
- Deployed the pipeline as a containerized service on Oracle Cloud Kubernetes, covering scheduling, execution, and output delivery
