Gaber Y.
AI Agent Architect & Engineer | RAG, LangGraph, MCP, LLM Apps
I build production AI agents, RAG systems and LLM apps, and fix the ones that break after the demo. Head of AI leading a 10-person team, 7 years in ML/AI, still hands-on in the code. New build, rescue of an existing system, or an extra senior engineer on your team. WHAT I DO • AI agents and multi-agent workflows: LangGraph, LangChain, MCP tool servers, tool use, human approval steps. • RAG and knowledge assistants that are measured, not guessed: evaluation sets, chunking, hybrid search, re-ranking, tracing, cost and latency per request. • LLM apps and chatbots on OpenAI, Claude and open models, including real-time voice AI. • Production hardening: guardrails, retries and fallbacks, tests in CI, Docker, Kubernetes, AWS and GCP. • Video and vision AI when the product needs it. SELECTED WORK • Enterprise multi-agent workflow platform (current role): led the architecture and hands-on build of the layer that turns visual workflow graphs into running agent workflows, with guardrails, agent coordination and tool access through a custom MCP server. • MarketScale (B2B video platform, Upwork contract): a conversational agent that turns videos into clips, podcasts and music; a QA agent that reviews edited videos; and "Channel Brain", a per-brand knowledge base built from videos, websites and calls on vector and graph databases. • Real-time vision systems: exam proctoring at Rosalyn.ai scaled to thousands of concurrent exams, and live MMA stream analytics at FX1. HOW I WORK • Start small: an architecture review or one working slice, so you see value before a bigger commitment. • Evaluation first: a set of real test cases and a baseline before any change, so every improvement shows up as a number. • I work in your repository, with a weekly call, short written updates, and honest calls on what to keep, fix or replace. STACK Python, LangGraph, LangChain, LangSmith, MCP, OpenAI, Anthropic Claude, Weaviate, Pinecone, pgvector, PostgreSQL, Redis, FastAPI, Docker, Kubernetes, AWS, GCP, PyTorch. Based in Dubai and Cairo (UTC+4 / UTC+2). MSc in Artificial Intelligence, University of Bath. Open to contract-to-hire. Tell me what you're building, or send one example that fails today, and I'll tell you where I'd look first.
Earnings and projects
Last 6 months ÷ 6
Last 6 months: $13k
Billed in all: $112,086$112k
$6.4k in the last 6 months
Across 20 contracts
Earnings, last 30 days
Read once a week before Oct 21, 2026, daily since. Dotted: longer than usual between two readings. Hatched: over two weeks without one.
Skill rankings: Top 7% in Prompt Engineering
Ranked by MRR among everyone on Upwork with the skill.
- #41LangGraph299 freelancers · 🇦🇪 #3Top 14%🇦🇪#3 of 7
- #46MLOps468 freelancers · 🇦🇪 #2Top 10%🇦🇪#2 of 7
- #65Model Context Protocol (MCP)413 freelancers · 🇦🇪 #3Top 16%🇦🇪#3 of 6
- #114Vector Database1,142 freelancers · 🇦🇪 #3Top 10%🇦🇪#3 of 15
- #188AI Model Integration1,618 freelancers · 🇦🇪 #5Top 12%🇦🇪#5 of 21
- #194Prompt Engineering2,859 freelancers · 🇦🇪 #2Top 7%🇦🇪#2 of 27
- #240AI Automation2,024 freelancers · 🇦🇪 #7Top 12%🇦🇪#7 of 26
- #254AI Consulting2,321 freelancers · 🇦🇪 #8Top 11%🇦🇪#8 of 40