Skip to content
Chapter I - Introduction

I build systems that read, understand,
& deliver knowledge.

Applied AI Engineer & Tech Lead crafting production RAG systems, interactive AI platforms, and full-stack applications. Never done learning. Always shipping.

Scroll
Chapter II - The Author

About Louay

I'm Louay Alshoum, a Senior Tech Lead and Applied AI Engineer at Verofax, where I've spent 4+ years shipping products across AI, AR, and full-stack development.

I led a production Arabic RAG system serving government clients across the GCC, with custom guardrails for regulated environments. I also built the AI backend for the Holobox, an interactive holographic avatar I've personally demoed at GITEX Global four years running (2022–2025), Biban 2024, and a major medical institution in Saudi Arabia.

Beyond building, I teach. I delivered 15 hours of courses on Neural Networks and Python Programming at Princess Nourah bint Abdulrahman University, and I regularly mentor engineers across RAG architecture, prompt engineering, and production deployment. Originally from Syria, I've called seven countries home from Damascus through Europe, Kuwait, China, Jordan, Lebanon, UAE, and Saudi Arabia.

My name means shield in Arabic. I like to think I protect the things that matter: knowledge, craft, and the curiosity to keep going.

By the numbers
Years of Experience0+
Projects Shipped0+
Countries Lived In0
Books Read This Year0
Chapter III - The Codex

Selected Works

Every project is a page in the codex, a record of what was built and what was learned.

01Featured Project
Production

Arabic RAG System

5+
Enterprise Clients
<1.5s
P95 Latency
2
Languages

Architected a production bilingual Arabic/English RAG system for government and enterprise clients. Focused on Arabic-specific NLP challenges including RTL text processing, diacritics handling, and dialect-aware retrieval. Designed caching and retrieval layers for low-latency responses in regulated environments.

PythonFastAPIRAGNLPVector Search
Tap to view →
02

Holobox AI Avatar Platform

Production

Built the AI backend for an interactive holographic avatar platform demoed at major events including GITEX Global (2022–2025), Biban 2024, and a major medical institution in Saudi Arabia. The platform delivers bilingual Arabic/English conversational interactions for government and enterprise clients across the GCC.

GITEX Global
3+
Major Events
3yr+
In Production
PythonNext.jsNLPAIEvals
Tap to view →
03

Smart Construction AI

In Progress

Building a RAG system for construction submittal management, enabling contractors and architects to query, track, and manage submittals and approval queues through natural language. Users can index project specs, query approved products and manufacturers, create auto-numbered submittals with revisions, and import existing logs from Excel.

4
Agent Modules
NLP
Interface
FAISS
Vector Store
PythonFastAPILangChainReactFAISS
04

Book Companion

In Progress

A multi-agent AI reading companion built with LangGraph. An orchestrator agent routes to specialized sub-agents — Ingestion, Analysis, and Synthesis — backed by a hybrid retrieval agent (query decomposition → hybrid search → rerank → merge) and a response agent that generates answers with citations and flashcards.

6
Agent Nodes
4-Stage
Retrieval Pipeline
LangGraph
Framework
PythonLangGraphMulti-AgentRAG
05

langchain-arabic

Published

An open-source LangChain output parser for Arabic text post-processing. Restores diacritics via a deterministic dictionary backend with an optional CATT neural fallback, and converts numbers to Arabic words across five contexts. Built as a BaseTransformOutputParser with streaming and async support out of the box.

<5ms
Dictionary Latency
5
Number Contexts
2
Dialect Support
PythonLangChainNLPOpen SourcePyPI
Tap to view →
Chapter IV - The Toolkit

Skills & Technologies

AI & ML

LangChainRAG SystemsAzure OpenAILangGraphVector Databases (PGVector)Azure AI SearchLangSmithOpen-Source LLMs (Qwen3)vLLMLLMOpsEvalsGuardrailsML Fundamentals

Backend

PythonFastAPINode.jsREST APIsPostgreSQLRedisMongoDB

Frontend

ReactNext.jsAngularTypeScriptTailwind CSSHTML/CSS

Tools & Infra

GitDockerKubernetesAzureAWSPrometheusCI/CDLinuxVercel
Chapter V - The Log

Learning in Public

Notes, reflections, and lessons from the journey. Never stop learning.

Why I Built langchain-arabic

AIMar 2026

LLMs generate Arabic fluently but strip the diacritics TTS needs to pronounce it. I built an open-source post-processing layer with benchmarks.

What Building an Arabic RAG System Taught Me

AIFeb 2026

The challenges of Arabic NLP, bilingual retrieval, and what changes when your RAG system needs to work in a right-to-left language.

From Angular to AI: Why I Switched Lanes

CareerJan 2026

After years of frontend work, I followed my curiosity into backend and AI. Here's what the transition actually looked like.

The Self-Taught Engineer's Dilemma

GrowthDec 2025

On imposter syndrome, AI tools, and why using every resource available makes you resourceful, not a fraud.

Chapter VI - Correspondence

Let's Build Something

Have a project in mind? Looking for an AI engineer who actually ships? Or just want to talk about books? I'd love to hear from you.