Ahmed R.

I build AI agents and automation systems that actually run in production — not demos. For the last 2 years I've shipped multi-agent systems, RAG pipelines, and LLM-powered internal tools, on top of 6 years of full-stack engineering. I currently lead an engineering team, so I build the way production requires: tested, monitored, documented, and maintainable by someone other than me. ── AI & AUTOMATION ── - Autonomous agents & multi-agent orchestration (Anthropic Agent SDK, tool use, planner/ reviewer/executor patterns) - RAG pipelines — chunking strategy, embeddings, pgvector, re-ranking, evaluation loops - MCP servers & custom tool integrations - Business workflow automation: document processing, data extraction, support deflection, reporting, lead qualification - AI chat on WhatsApp Business API, web, and Slack — with persistent memory and real business logic behind it - Cost & latency engineering: caching, model routing, token budgeting ── FULL STACK ── - Backend: Laravel, Node.js, Python (FastAPI / Django) - Frontend: Next.js, React, Vue, TypeScript, Tailwind, Inertia - Data: PostgreSQL + pgvector, MySQL, Redis, Supabase, MongoDB - Architecture: multi-tenant SaaS, queues & background jobs, REST APIs, webhooks ── CLOUD & DEVOPS ── - AWS (EC2, S3, RDS, Lambda), Docker, CI/CD, Nginx, Linux - Laravel Forge, zero-downtime deploys, monitoring & alerting ── INTEGRATIONS ── - Payments: Stripe, Paymob, Checkout, Tabby, Tamara - Commerce: Salla, Shopify, WooCommerce - Comms: WhatsApp Business API, Twilio, SendGrid - Maps: Google Maps, Mapbox ── RECENT WORK ── - Multi-agent system where specialized agents plan, execute, and review each other's output before anything ships - Multi-merchant WhatsApp automation platform — AI replies, order flows, merchant dashboards - RAG assistant over private company documentation, answering with source citations - Production SaaS apps on the Salla marketplace (OAuth, webhooks, multi-tenancy) ── HOW I WORK ── Short call to understand the actual problem → written scope with milestones → a working slice early, not at the end → weekly updates. No black box. If you have a process eating hours every week, message me describing what it looks like today. I'll tell you straight whether AI is the right fix — or whether a simpler solution would serve you better.