End-to-End AI Engineer
I design & build AI systems from architecture to deployment.
From agents that qualify leads to football intelligence and real-time Arabic voice experiences, I own the full stack — architecture, integrations, AI pipeline, deployment, and operations. Tool-grounded, resilient, and built for technical conversations.
- InputChat message, voice call, webhook
- ReasonLLM routing, retrieval, tool calls
- ActReply, book, update records
Selected work
Projects built for a technical conversation.
Each page is a practical map of the system: its problem, architecture, ownership, trade-offs, and implementation details.
AI Broker
Built end-to-end multi-channel lead qualification agent with LLM routing, tool calling, real-time conversations, appointments, and operational controls.
CoachAI Chat
Built end-to-end evidence-grounded RAG system for tactical analysis, coaching reports, Arabic translation, and football decision support.
Arabic Voice AI
Built and deployed end-to-end real-time Arabic voice pipeline (STT → LLM → TTS) with telephony-aware processing and operational controls.
Technical stack
Python, FastAPI, Flask, REST APIs, WebSockets, async workflows
LLMs, RAG, embeddings, structured retrieval, evaluation pipelines
React, TypeScript, Docker, MongoDB, SQLite, Redis, Prometheus
WhatsApp, Calendly, OpenAI, Claude, Azure Speech, ElevenLabs, VoIP
The interview lens
These are not project summaries. They are evidence trails.
Each featured system shows the flow of requests, state, external dependencies, and failure boundaries.
Each claim is separated into my direct contribution and components I integrated or extended, so the implementation can be discussed honestly and in depth.