Final-year AI & Computer Science student at the University of Birmingham Dubai
Building reliable backend systems, applied-AI products, and decision tools in the UAE.

I enjoy engineering products where software, data, and real decisions meet. My work spans secured Java services, Python machine-learning pipelines, full-stack TypeScript applications, deterministic optimization, data contracts, and AI systems with strict structured outputs.

The common thread is practical: understand the domain, make the recommendation explainable, protect the data boundary, and prove the system works with tests rather than hype.

• Currently: developing FPL Intelligence, an ML-focused analytics project that turns football data into explainable decisions.
• Interested in: backend engineering, applied AI/ML, fintech, data products, and technology delivery.
• Based in: Dubai, United Arab Emirates.

Recommendation-first Fantasy Premier League intelligence for the live 2026/27 season.

Built a rules-versioned decision platform that combines multi-Gameweek projections, player availability, squad optimization, transfers, captaincy, chips, and historical simulation—then presents one primary recommendation with safe and aggressive alternatives.

• Point-in-time-safe modelling across multiple FPL rule eras.
• Deterministic planners for squad, bench, captain, transfer, and chip decisions.
• Immutable data snapshots, rules hashes, leakage safeguards, and persisted evidence.
• 304 Python tests, 32 frontend unit tests, and 32 desktop/mobile browser tests at the latest verified baseline.

An AI decision workspace that turns messy conversations into ranked decisions, visible trade-offs, and a useful next question.

Designed the product around the moment before a decision: users paste rough conversations or perspectives, and Friction extracts the real decision, scores what matters, compares options, classifies the tension, and creates a shareable brief.

• Strict model-output contracts validated with Zod.
• Server-side API-key boundary with safe local fallback analysis.
• Saved decisions, templates, journal notes, sharing, printing, and PDF briefs.
• Privacy-first local storage and resilient responsive workflows.

A secured wealth-management platform for advisor-led suitability, portfolio drift, transaction review, and rebalancing.

• JWT authentication, advisor ownership checks, and protected REST APIs.
• PostgreSQL persistence, Flyway migrations, immutable transaction ledger, and audit logs.
• Risk scoring, suitability rules, allocation drift, and rebalancing recommendations.
• JUnit, security, and Testcontainers coverage with GitHub Actions CI.

A responsive calorie and macronutrient tracker built from REST API to browser UI.

• Food logging, custom foods, serving-size recalculation, and daily macro totals.
• TDEE calculation, user targets, responsive charts, and dynamic client-side state.
• Spring Data JPA, Spring Security/JWT structure, Docker, Render, and CI.

Live demo

I contribute to established Java projects and work through real maintainer feedback, repository conventions, regression tests, and CI—not isolated coding exercises. I aim to pair each behavior change with focused tests and clear review notes so maintainers can validate it quickly.

Completed an internship supporting Panorama 2.0 AI integration and built a privacy-safe local enterprise KPI dashboard with a Python server, metric drill-downs, charts, alerts, evidence lineage, data APIs, and print-ready reporting. The portfolio-safe edition deliberately uses synthetic data and keeps private operational sources outside Git.

Worked across fintech market research, financial modelling, and GCC wealth-management strategy—experience that directly informed the domain thinking behind MandateIQ.

• Deloitte Australia Data Analytics: Tableau dashboarding, Excel classification, and forensic-technology analysis.
• Mastercard Cybersecurity: phishing-threat identification, security-awareness analysis, and targeted training recommendations.

Understand the decision → model the domain → protect the boundaries
→ test the failure modes → measure the outcome → communicate it clearly

• Evidence over hype: a model is useful only when it improves the real decision.
• Security and privacy by design: credentials stay server-side and sensitive data stays outside public repositories.
• Production-minded prototypes: authentication, migrations, CI, observability, fallbacks, and documentation matter.
• Readable systems: code, interfaces, and explanations should make trade-offs visible.

Open to software engineering, backend, applied-AI, fintech, and technology-delivery opportunities.

moustafa.ameen.useful@gmail.com · LinkedIn · Repositories