Case study
Masar X
Bilingual AI study platform: semantic PDF search and vector document indexing across web, desktop, and mobile clients on a single Supabase backend.
- Next.js
- TypeScript
- Supabase
- pgvector
- Electron
- Expo
Ahmed Aboalayoun · Full-Stack AI Engineer
Production and open-source builds, each with its evidence on the page.
Case study
Bilingual AI study platform: semantic PDF search and vector document indexing across web, desktop, and mobile clients on a single Supabase backend.
Case study
Open-source CLI (28 stars) that resets device and machine identifiers for AI coding IDEs, with guided flows, audit logs and restore scripts on Windows and Linux. Positioned as developer privacy hygiene; the project's own docs tell users to comply with each vendor's terms of service.
Case study
Production commerce and ordering web application, built and operated under a deterministic AI software-factory pipeline: locked specifications, milestone gates and CI-enforced quality checks. The source is private, so every statement on this page is owner-attested rather than publicly verifiable.
What I build for clients, from approved spec to handover.
A production web application or dashboard — not a prototype: real data, user accounts with roles, and the views and reports your team uses daily. Deployed, documented, and handed over to you.
All servicesA tool that turns your documents into answers: semantic search and Q&A over your own files — PDFs, contracts, knowledge bases — with source citations and access control, instead of manual digging.
All servicesAutomations and agent tooling that connect the systems you already use: MCP servers, scheduled jobs, data sync between systems, and deterministic CI pipelines that make every change testable and reversible.
All servicesA deterministic factory, not improvised code.
Every build starts from a short written spec you approve: scope, constraints, and what "done" means, before any code exists.
Work ships in thin, working slices you can open and use, not a big reveal at the end.
Lint, types, tests and build run automatically on every change. Nothing merges without passing.
You get documentation, deployment and the keys: a system you own and understand.
Public record, verified and dated. Statuses stay honest: merged stays merged, open stays open, acknowledged stays acknowledged.
Config-CLI hardening merged into hermes-agent main (2026-09-18). Includes the salvaged fix from PR #109834 (@fotedev) — credited in the PR body as "source hunk, earliest filer" — with an attributed commit; authorship preserved.
Verified on 2026-11-06
View on GitHubOpen pull request (last updated 2026-10-17). Status: open — not merged.
Verified on 2026-11-06
View on GitHubOpen pull request (last updated 2026-10-17). Status: open — not merged.
Verified on 2026-11-06
View on GitHubResponsible security disclosure. Acknowledged by the vendor on 2026-09-21 ("Acknowledged on both counts… Forwarded to security review"). Issue closed 2026-10-19 as not_planned — no fix PR referenced; never labeled "fixed".
Verified on 2026-11-06
View on GitHubCertificate
Dubai Future Foundation · Dubai Centre for Artificial Intelligence
Completion of the One Million Prompters initiative — a prompt-engineering training program. The certificate credits Dubai Future Foundation and the Dubai Centre for Artificial Intelligence. Program page: dub.ai/omp.
Recipient: Ahmed Aboalyoun
The certificate PDF opens outside this site.
I'm Ahmed Aboalayoun (fotedev), a Full-Stack AI Engineer. I build web platforms, production RAG systems and the tooling around them, and I run my own work through a spec-gated software factory so every claim on this site can be traced to evidence.
More about meEmail is the fastest way to reach me — or send the form on the contact page. Either way it lands straight in my inbox.