SYSTEM ONLINEOpen to opportunities

AbdulrahmanAlanani

AI Engineer

Generative AI EngineerLLM EngineerComputer Vision Engineer

I build intelligent systems that connect AI models with real-world software, automation, data, and hardware.

From model to production. From software to the physical world.

Explore the Work Download CV Open a Channel

AI
LLMs
RAG
Computer Vision
Automation
Robotics

↓ Scroll to initialize

01 / IDENTITY

An engineer who ships intelligent systems

I'm Abdulrahman Alanani, an AI Engineer based in Riyadh, Saudi Arabia. My work sits at the intersection of machine learning, generative AI, backend engineering, and physical hardware — I take models out of notebooks and wire them into systems that actually run.

Since graduating with a B.Eng. in Computer & Systems Engineering (Badr University in Cairo, 2024), I've built an anomaly detection system for IT infrastructure monitoring, a YOLOv8-powered fire detection drone graded A+, an end-to-end AI integration platform spanning RFQ automation, quotation APIs, and bilingual RAG — and I'm currently going deep on LLM engineering.

I care about measurable outcomes: downtime reduced ~15%, incident resolution ~20% faster, retrieval systems that refuse to hallucinate. If it can't be tested, measured, or deployed, it isn't finished.

“From model to production. From software to the physical world.”

EXPERIENCE

AI/ML Systems Support Engineer

Dec 2024 – Dec 2025 · Cairo

Problem Solver Org. / Future of Egypt

Built and deployed an anomaly detection model for IT infrastructure monitoring, contributing to ~15% downtime reduction.
Automated diagnostics and troubleshooting workflows in Python, improving incident resolution speed by ~20%.
Maintained ~95% first-contact resolution across 200+ supported incidents per month.

How I Operate

Complex problem decomposition
Data-driven decisions
Delivering on commitments
Independent troubleshooting
Software × hardware fluency
Relentless learning curve

02 / CAPABILITY MAP

Engineering Intelligence Map

Not a wall of logos — every node links to the projects where it was actually used. Hover to illuminate connections, click any skill for evidence.

Hover to explore · Click a node for proof of work

AI / ML

Generative AI

Engineering

Automation

Robotics & Edge

03 / PROJECT LAB

Systems, not screenshots

Each flagship project is presented the way it was engineered: as an architecture you can walk through.

FLAGSHIP // 01View repository

AI Integration Platform

RFQ → CRM Automation · Quotation API · Bilingual RAG

An end-to-end AI integration assessment for AL ROUF LED Lighting: three runnable subsystems covering document-understanding automation, a production-style FastAPI microservice, and a citation-enforced bilingual RAG service — all containerized with Docker.

ROLE

System architecture · API development · AI integration · Workflow automation · Testing

STACK

Python · OpenAI · n8n · FastAPI · SQLite · Docker · pytest

OUTCOME

3 subsystems delivered · automated tests · Docker packaging · OpenAPI docs · refusal logic against hallucination

LIVE ARTIFACT // n8n workflow canvas — Task 1: RFQ → CRM

LIVE ARTIFACT // OpenAPI docs — Task 2: Quotation service

Select subsystem

RFQ Webhook

Inbound RFQ documents arrive through an n8n webhook endpoint, triggering the automation on every submission.

◦ Click any node to inspect it

FLAGSHIP // 02

Autonomous Fire Detection & Extinguishing DroneGraduation Project — Grade A+

A six-person graduation project: an autonomous quadcopter that detects fire in real time with a custom-trained YOLOv8 model running on a Raspberry Pi 3, integrated with an APM 2.8 / ArduCopter flight stack and engineered around an AFO fire-extinguishing-ball payload — full path from dataset preparation to airborne inference and PS4 ground control.

◉ MISSION CONTROL

LIVE TELEMETRY

AI Vision

Custom-trained YOLOv8 model detecting fire in real time (mAP@50 ≈ 0.824 · Precision ≈ 0.83), trained through data collection, processing and model training on ~754 validation images.

≈ 0.824

mAP@50

≈ 0.83

Precision

~754

Validation images

A+

Grade

NOTE // Suppression ball designed but not procured — vendor restricted sales to institutions. Documented honestly in the project report.

Hardware

Raspberry Pi 3 — onboard compute hub
APM 2.8 · ArduCopter autopilot
4× 2200kV brushless motors + PWM ESCs
LiPo battery · power distribution · voltage regulator
Aluminum frame · Raspberry Pi camera

Software

Python · YOLOv8 · OpenCV
Mission Planner — motor/ESC/compass calibration & ground station
Pygame + pySerial — PS4 control link to APM 2.8
Real-time detection alerts with bounding boxes

DETECTION CHAIN

Camera Feed

Onboard camera captures live video during autonomous flight over target areas.

◦ Click any node to inspect it

More builds

LLM Engineering Journey

IN PROGRESS

A structured, hands-on path into production LLM engineering — working tools, not notes. Includes an AI job-description analyzer with structured JSON output and a dual-backend technical tutor streaming from both OpenAI and local Ollama models simultaneously.

Streaming responses across cloud and local LLMs
Structured JSON output with defensive parsing
Expert system-prompt design
Roadmap: RAG pipeline, agents, fine-tuning, production deployment

View repository

ML Internship Project

Four end-to-end classical ML tasks executed and evaluated in Google Colab — regression, NLP classification, churn prediction, and time-series analysis — with trained model artifacts persisted for reuse.

Verified results

LIVE ARTIFACT // churn evaluation — held-out test report

0.66

House price R²

0.80

Sentiment accuracy

0.95

Churn accuracy

0.886

Churn ROC AUC

House price regression (California Housing): R² ≈ 0.66
Sentiment classification: TF-IDF → neural network, accuracy ≈ 0.80
Churn prediction: Random Forest, accuracy ≈ 0.95 · ROC AUC ≈ 0.886
Class-imbalance handling via oversampling

04 / METHODOLOGY

How I think

Not a slogan — the actual loop I run on every problem, mapped to real moments from my projects. Click any stage.

STAGE 01 / 8

Problem Detected

Spot the signal in noise — define what actually broke before touching anything.

APPLIED IN REALITY // واقعياً

200+ monthly IT incidents hid recurring failure patterns no one had quantified.

◦ Click a stage to see it applied

05 / CONCEPT LAB

Build with me

Pick a problem and watch how I would architect the solution — the same thinking that powered the projects above.

What do you want to build?

Not limited to these — I can build anything a software engineer or an AI engineer can, and I'm always up for new tools and unexplored problems.

How I would approach this problem:

RAG System

Documents

Ingest domain knowledge in any language (AR/EN).

◦ Click any node to inspect it

06 / JOURNEY & SIGNALS

An engineering trajectory, measured

Every step here produced something verifiable — a deployed system, a graded project, a measured metric.

ENGINEERING SIGNALS

Major systems built

Subsystems in one platform

Downtime reduction contributed

Faster incident resolution

First-contact resolution

Incidents supported / month

2018 – 2024

Computer & Systems Engineering

B.Eng. at Badr University in Cairo — software, embedded systems, and the foundations of intelligent systems.

B.Eng.
Embedded
Systems

2024

Fire Detection Drone — Grade A+

Led CV + flight integration for a six-person graduation project: YOLOv8 onboard a Raspberry Pi, mAP@50 ≈ 0.824.

YOLOv8
Raspberry Pi
APM 2.8

Dec 2024 – Dec 2025

AI/ML Systems Support Engineer

Problem Solver Org. / Future of Egypt — anomaly detection for IT infrastructure (~15% downtime cut), automated diagnostics (~20% faster resolution), 95% FCR across 200+ incidents/month.

Anomaly Detection
Python
Automation

2025

AL ROUF AI Integration Platform

Three production-style subsystems delivered: RFQ→CRM automation, quotation microservice, bilingual RAG with refusal logic.

n8n
FastAPI
RAG
Docker

CERTIFICATIONS

LLMOps Specialization

DeepLearning.AI

LLM Engineering — Full Stack AI ApplicationsLIVE

Ongoing program

OCI 2025 AI Foundations Associate

Oracle

Machine Learning Foundations

AWS Educate

Introducing Generative AI with AWS

Udacity

Artificial Intelligence Diploma

ITI Cairo

07 / SYSTEM TRANSPARENCY

How this portfolio works

The site itself is a working RAG system — the same architecture Abdulrahman builds for clients, applied to his own portfolio. Ask the assistant anything and watch the pipeline:

Knowledge chunks indexed

817

Vocabulary terms

< 5 ms

Typical retrieval time

Visitor Question

You type a question in Arabic or English and pick an answer style: Recruiter, Engineer, or Student.

◦ Click any node to inspect it

08 / COMMUNICATION CHANNEL

Let's build something intelligent.

Have a challenging problem, an AI idea, or an opportunity worth exploring? Open a channel.

aalanani.2000@gmail.com

in/abdulrahman-alanani

GitHub

github.com/aalanani2000

Open to roles

LLM Engineering
AI Integration
Computer Vision
AI Automation

Download CV

Open to opportunities·Riyadh, Saudi Arabia · Open to remote·Usually responds within 24 hours