AI & full-stack engineer · Python, FastAPI, RAG, React / Next.js · Open to remote & relocation
I build reasoning systems, animated websites, RAG pipelines and field tools, then the screens you meet them on.
I learned to trace a packet before I learned to trace a prompt.
Turned "listings take too long to process" into a Google Forms → Sheets → Apps Script pipeline that agents adopted on day one, plus a listing filter and an automated PDF brochure generator. The only technical person on the engagement, so network, hardware and backups were mine too.
Shipped 5+ production client websites on deadline and improved page-load speed by about 40%, using AI tools across the build and coordinating stakeholders in Lark and Trello.
Hardware diagnostics with a multimeter, lab connectivity faults traced to the switch, Windows imaging with Macrium Reflect and Sysprep, and root causes written down.
LIU graduate. Cisco Networking Academy: CCNA Switching, Routing & Wireless Essentials; CCNA Enterprise Networking, Security & Automation; Network Security Specialist.
I wrote field-tools: dependency-free scripts for industrial gateways, including a subnet scanner that looks for Modbus TCP on port 502. grep · ping · tar
Retrieval, multi-agent orchestration and a neuro-symbolic mind of my own. Software that thinks, and can show why.
Motion with a job, type with a voice, and a CMS the client actually wants to use. Builds marked Concept build were designed and built as speculative proposals; they were not commissioned.
Simple, beautiful sites for restaurants, cafés, clinics and shops in Beirut. The studio site is itself the demo.
Visit the studio ↗The whole studio in one repo: a white-label demo builder, a lead scraper, a bilingual AR/EN brochure and the outreach kit.
See the repo ↗
Floating luxury for an interior-architecture studio: nature, light, scenography.
Visit the site ↗
Ocean Breeze: calm-cinematic motion and a CMS staff can run alone.
Visit the site ↗
Paste a messy URL. Get a clean, short link. No accounts, no tracking.
Visit the site ↗Six projects that show how I work. Every card carries one honest status.
A local neuro-symbolic mind. A symbolic kernel reasons, three stores remember, a learning loop compiles proofs into rules, and a small language model gives it a voice.
python -m interface.cli --webRead the case study →02A designed, not-yet-built LangGraph team of specialist agents meant to turn a Next.js + Payload + GSAP template into a finished, production-grade site.
python -m site_builder --spec site.jsonRead the case study →03Ask questions about IT security policies and get answers with section citations. Hybrid retrieval, RRF fusion and a cross-encoder reranker, served by FastAPI.
qdrant --hybrid --rrf --rerankRead the case study →04An AI-assisted vulnerability scanner for Python codebases. Bandit, Safety and Semgrep find the issues; an LLM explains the fix side by side with the vulnerable code.
bandit && safety && semgrep && groq --fixRead the case study →05A rebuild of a travel agency's website with a calm-cinematic motion system and a CMS staff can run without a developer. Designed and built as a speculative proposal.
next devRead the case study →06A decision ladder for AI coding agents. Climb eight rungs, stop at the first one that holds, then run a discipline gate on the diff. It catches over-engineering before it ships.
chisel --ladder --gateRead the case study →A browser message becomes an intent, a memory query, a proof, and a voice.
The chisel ladder: climb in order and stop at the first rung that holds. It's how this site was built too: one animation engine, no framework runtime.
Three roles, six projects, the full skill set and the Cisco CCNA-track certificates, on one page that applicant-tracking systems can read.