AI Engineer · Gaza, Palestine · Remote
At OSOS, I work on GAITH, developing AI features and workflows, researching technical options, planning implementation, reviewing changes, and debugging issues across the platform. My software-verification background shapes how I test and ship AI-enabled systems.
current work
I work on GAITH, OSOS's AI platform. My work spans AI workflows and knowledge systems, backend reliability, platform integrations and identity, and the engineering work needed to move changes from research through testing and release.
LLM-enabled workflows, document ingestion and retrieval, RAG behavior, and the failure cases around indexing and grounded answers.
Execution and resume behavior, retries, worker and file-processing failures, observability, and regression safeguards.
OAuth/OIDC, token lifecycle, webhooks and subscriptions, account scoping, and integration work across platforms such as Microsoft 365, Google Workspace, Atlassian, and GitHub.
Technical research, solution design, Jira task breakdown, pull-request review, testing, rollout checks, and production debugging.
selected work
Selected projects across AI engineering, backend systems, developer tooling, and software products.
Repair Shop Management System
A bilingual Arabic/English system for repair shops to manage devices from intake to delivery.
RoleProduct planning, research, architecture, implementation, verification, deployment, and documentation
Repair workflows · Arabic/English · RTL · Self-hosted · Verification
Human gates
Agent tools
Claude Code · Codex · MCP · Skills · Hooks · Worktrees
Additive framework around GitHub Spec Kit (not an official fork)Spec-driven AI Engineering Framework
A framework that turns long AI coding sessions into planned, reviewable, test-backed engineering work.
RoleDesign and implementation
Claude Code · Codex · MCP · Worktrees · Review gates
Runtime stack
Windows packaging
Arabic RTL · Bundled JRE · Versioned migrations
Offline Arabic Point of Sale
An offline-first Arabic desktop POS built around reliable local data, inventory, sales, and repeatable Windows packaging.
RoleArchitecture and implementation
Offline-first · Arabic RTL · Desktop · SQLite · Packaging
76% accuracy
Small local dataset96% accuracy
Larger dataset, more featuresDifferent datasets — not directly comparable
Machine-learning models that ranked lighting-pole maintenance priorities from condition data.
RoleSoftware Engineer
Machine learning · Data cleaning · Model comparison · RapidMiner
experience
Jul 2025 – Present
Remote
I work on GAITH, OSOS's AI platform. My role covers feature development, technical research, planning, debugging, code review, testing, and release support.
Aug 2021 – Oct 2023
Gaza, Palestine
Worked on software verification for NVIDIA firmware tools used with switches and network interface cards. Maintained Python regression tests, investigated failures, tested releases, wrote test plans, and worked with verification, QA, and engineering teams.
Aug 2021 – Jan 2022
Gaza, Palestine
Taught Data Structures and Algorithms in Java to about 90 students, covering core data structures, recursion, and time and space complexity.
approach
I keep the engineering process explicit: understand the system, plan the change, implement with review, and stay with it through debugging.
01
I read the existing system and compare technical options before committing to a design. The current code and documented behavior are the starting point.
02
I turn decisions into scoped tasks, clarify requirements, and keep implementation and review work easy to follow.
03
I use Claude Code, Codex, MCP, skills, hooks, memory, and worktrees, but I keep tests, review, and acceptance criteria outside the model's judgment.
04
I trace issues across services, logs, state, and integrations, then add tests or safeguards so the same failure is easier to catch next time.
skills
LLM applications, RAG, embeddings, document ingestion, knowledge-base indexing, and agentic workflows in production contexts.
LLM applications · RAG · Embeddings · Document ingestion
Python and Java are my primary languages. I build backend services with FastAPI, Celery, Redis, PostgreSQL/pgvector, and SQLite, with working experience in TypeScript, JavaScript, Go, and PHP.
Python · Java · FastAPI · PostgreSQL · Celery
Claude Code, Codex, MCP, subagents, custom skills, hooks, spec-driven development, review gates, worktrees, and parallel agent workflows.
Claude Code · Codex · MCP · Skills · Review gates
OAuth/OIDC, REST APIs, webhooks, subscriptions, token refresh, account scoping, and multitenant access patterns across enterprise integrations.
Microsoft 365 · Google Workspace · Atlassian · GitHub
Test planning, automation, regression testing, manual testing, unit testing, and behavioral checks shaped by a verification background.
pytest · Playwright · Regression testing · Behavioral checks
Git, Linux, Bash, Docker, Docker Compose, GitHub Actions, CI/CD, Kubernetes, Helm, and GCP.
Docker · GitHub Actions · CI/CD · Kubernetes · GCP
background
I moved from software verification into production AI and backend engineering. I prefer work that starts as an unclear problem and ends as a tested, maintainable system.
My verification background still shapes how I approach changes. I research the existing behavior, make the change explicit, test the failure paths, and stay with issues through debugging.