Skip to content
Seeking a 6-month final-year internship — January/February 2027

Abderrahman
Kayouh.

>_

Engineering student in Data & AI, I bridge applied research and production by designing intelligent systems combining predictive Machine Learning, data engineering, and LLM-based solutions.

Morocco · Open to relocation
2+
Internships
1
R&D projects
7
Team Led
Abderrahman Kayouh

// 01. about_me

About me

My path has been shaped by a single ambition: to understand data, design intelligent systems, and turn them into concrete solutions.

Across my projects and experiences, I have tackled a wide range of problems — from data engineering and Machine Learning to computer vision, LLMs, and automation. These experiences taught me to look beyond the model or the algorithm: to understand the need, build a robust solution, measure its performance, and think through its deployment.

In parallel, my experience in a Junior Enterprise — notably leading a team of 7 — helped me grow my leadership, autonomy, and ability to collaborate on concrete projects.

Today, I want to put this blend of technical and human skills at the service of ambitious projects, where I can keep learning, building, and creating value.

ArabicNativeFrenchFluentEnglishProfessional
whoami.json
name
: Abderrahman Kayouh
role
: Engineering Student — AI / ML & Data
location
: Morocco · Open to relocation
email
: abderrahmankayouh67@gmail.com
status
Open to final-year internships in ML / AI / Data Engineering & research
0+

Marine data points analysed — NASA predictive ML (Global Nominee)

0

Professional certifications earned

0+

Projects across ML, data & full-stack

0

Engineers led as JEEA President

// 02. skills

Technical skills

The stack behind the work — from model training to data pipelines to production.

machine_learning.py

Machine Learning & AI

  • ML Modeling with Scikit-learn (Classification & Regression)
  • Deep Learning Basics (TensorFlow, Keras)
  • Predictive Modeling & Feature Engineering
  • Model Evaluation, Validation & Metrics Optimization
  • NLP & Conversational AI (Foundations)
  • Applied RAG & Document Understanding
  • Computer Vision & Applied OCR
data_engineering.sql

Data Engineering & BI

  • SQL & Oracle PL/SQL
  • ETL Pipeline Design & Data Cleaning
  • Data Warehouse Modeling (Star Schema)
  • Business Intelligence (Power BI, Tableau)
  • Relational & Cloud Databases (PostgreSQL, Supabase)
  • Decision Reporting & KPI Tracking
mlops.yaml

MLOps, Tools & Delivery

  • Model Serving & REST APIs (FastAPI / Flask)
  • Basic Containerization (Docker)
  • Workflow Automation (n8n, Make)
  • Version Control (Git & GitHub)
  • Agile & Scrum Delivery / Team Coordination
languages.sh

Languages & Technologies

  • Python (Pandas, NumPy, Scikit-learn, TensorFlow/Keras)
  • SQL (PostgreSQL, MySQL, Oracle)
  • Java / Spring Boot (Beginner / Academic)
  • TypeScript / React (Beginner / UI Basics)
  • Bash & Linux (Basic Commands & Scripting)

// research_interests

Research Interests

Directions I want to explore further during a research-oriented PFE.

Retrieval-Augmented Generation & LLM SystemsMultilingual NLP & Document AIComputer Vision & OCRContinual / Personalized LearningData & ML Systems at Scale

// 03. projects

Selected work

Case studies from production ML to independent research — filtered by domain.

Engineering Projects

Production-oriented systems — end-to-end delivery, deployment and measurable impact.

Vala Bleu Ops Copilot — Agentic RAG & MLOps
NLP & LLMIn progressPFA · Internship — Vala Bleu, Agadir · In progress

Vala Bleu Ops Copilot — Agentic RAG & MLOps

An operational copilot for IT Ops at Vala Bleu (web host, Agadir): multilingual FR/EN document RAG and server-log anomaly detection, orchestrated by an agentic router, with 100% local LLM inference on an NVIDIA GPU.

Solves · Unifying business knowledge (docs, tickets) and operational state (logs) in one assistant, without letting sensitive data leave the premises.
  • Multilingual FR/EN document RAG (PostgreSQL + pgvector).
  • Server-log anomaly detection (Isolation Forest).
  • Agentic router directing queries to RAG, Logs, or both.
  • 100% local LLM inference on NVIDIA GPU; sensitive-data anonymization.
  • RAGAS evaluation; FastAPI API + Streamlit dashboard, dockerized.

// stack

PythonFastAPILangChainpgvectorRAGASDocker
TomatoSort Pro — Real-Time Tomato Sorting (CV)TomatoSort Pro — Real-Time Tomato Sorting (CV)TomatoSort Pro — Real-Time Tomato Sorting (CV)TomatoSort Pro — Real-Time Tomato Sorting (CV)
Machine LearningComputer VisionPersonal · Computer Vision · 2025

TomatoSort Pro — Real-Time Tomato Sorting (CV)

A real-time computer-vision system that sorts cherry tomatoes on a line — flagging healthy tomatoes vs. foreign objects from a live camera feed, CPU-optimized for edge deployment.

Solves · Manual, slow and error-prone visual quality inspection on a tomato sorting line.
  • YOLOv8-Nano detection (tomato / foreign_object) — mAP@50 0.846, precision 0.92, recall 0.86.
  • Intel OpenVINO CPU-optimized inference for real-time edge deployment.
  • Streamlit dashboard + lightweight OpenCV window, with MLOps & PLC-control scaffolding.

// stack

YOLOv8OpenVINOOpenCVStreamlitPython
AutoExpert — Automotive Agency PlatformLiveOpen
Full-StackInternship — AH Digital · 2025

AutoExpert — Automotive Agency Platform

A web platform for an automotive agency with an automated backend (n8n) and real-time reporting (Supabase) over 50+ vehicles.

Solves · Manual vehicle-catalogue management and slow, disconnected reporting for an auto agency.
  • Automated backend workflows using n8n.
  • Managed a catalogue of 50+ vehicles with real-time Supabase reporting.

// stack

n8nSupabaseAutomationFull-Stack
Hotel Management System — Full-Stack
Full-StackUniversity project · 2025

Hotel Management System — Full-Stack

A full-stack hotel management system (Java / Spring Boot / React / Oracle PL/SQL) handling 200+ reservations across 5 business modules.

Solves · Fragmented, manual handling of 200+ hotel reservations across disconnected processes.
  • 5 business modules with Oracle PL/SQL procedures.
  • Java backend exposing REST APIs; React frontend.
  • Managed 200+ reservations with reporting via REST.

// stack

JavaSpring BootReactOracle PL/SQL
Supply Chain BI — DataCo Analysis
Data EngineeringUniversity project · 2025

Supply Chain BI — DataCo Analysis

A business-intelligence project over 100,000+ rows: ETL into a star-schema warehouse feeding 8 financial KPIs and interactive Power BI dashboards.

Solves · Turning 100k+ rows of raw supply-chain data into decision-ready financial KPIs.
  • ETL over 100,000+ rows.
  • 3-star-schema data warehouse.
  • 8 financial KPIs surfaced through interactive Power BI dashboards.

// stack

ETLData WarehousePower BIBI
AH-Chat-Digital — Support Automation & NLP Integration
AI & AutomationProductionInternship — AH Digital, Technopark Agadir · 2025

AH-Chat-Digital — Support Automation & NLP Integration

A customer support automation dashboard integrating n8n workflows and NLP for automated inquiry routing, reducing response times by ~60%.

Solves · Repetitive support inquiries causing long response times and manual dispatching overhead.
  • Configured end-to-end automation pipelines using n8n to connect chat channels with internal workflows.
  • Integrated lightweight NLP and prompt parsing to classify incoming messages and draft automated replies.
  • Built an interactive tracking dashboard to monitor conversation status and reduce response bottlenecks by ~60%.

// stack

n8nWorkflow AutomationNLP BasicsDashboardAPIs

Research & Experimental Work

Experimentation, methodology, evaluation and technical challenges — early-stage, exploratory research directions.

Research Project
R&DIn progressIndependent research project · Document AI · In progress

Research Project

Reliability-Aware OCR-Free Extraction of Structured Data from Degraded Handwritten Medical Prescriptions — a fine-tuned Donut vision-language transformer (Swin encoder, mBART decoder) that maps prescription images directly to structured JSON, trained entirely on synthetic data, with decoder-confidence selective prediction so it knows when it doesn't know.

Solves · Handwritten Moroccan prescriptions (FR/AR) hold sensitive data that can't be freely collected, yet a deployable clinical extractor must know when its reading is wrong.
  • OCR-free pipeline: a fine-tuned Donut (Swin encoder + mBART decoder) maps the raw image directly to structured JSON — no intermediate OCR, no error propagation.
  • A fully synthetic generator injects realistic physical degradations (projective warp, ink bleed, uneven lighting, sensor noise, stamps) with per-character handwriting jitter to bridge the sim-to-real gap.
  • 89.4% macro exact-match and 0.922 medication-list F1 on 200 held-out images — ~80 points over a classical modular OCR baseline (87.9% vs 6.1% macro EM).
  • Decoder-confidence selective prediction lifts field precision from 88.0% to 98.6% at 83.5% coverage; attention is well-grounded but does not flag hallucination — confidence does.
  • All data are synthetic; a research proof-of-concept, not a validated medical device.

// stack

Document AIOCR-FreeDonut / VLMSynthetic DataSelective Prediction
NASA Space Apps — Smart Shark Behaviour TagLiveOpen
Machine LearningGlobal NomineeNASA Space Apps Challenge — Global Nominee · 2025

NASA Space Apps — Smart Shark Behaviour Tag

A NASA Space Apps Challenge project — Global Nominee — applying predictive machine learning to real-world marine data to track apex-predator behaviour in real time.

Solves · Turning noisy, real-world marine environmental data into real-time behavioural insight on apex predators.
  • Global Nominee — earned international recognition at the NASA Space Apps Challenge.
  • Predictive ML & analytics on 10,000+ real-world marine data points; surfaced migratory patterns.
  • Designed a smart-tag concept and shipped an end-to-end prototype under hackathon pressure.
  • Delivered through tight team collaboration and rapid, real-world problem-solving.

// stack

AI & MLPredictive AnalyticsMarine DataTeamwork
ALM Surrender Risk
Machine LearningIn progressActuarial & Finance · In progress

ALM Surrender Risk

Life-insurance surrender-risk modelling: a Monte-Carlo simulation engine, interest-rate calibration (Vasicek / Hull-White) and a predictive ML model with SHAP, served through a FastAPI API and a Streamlit dashboard under a Solvency II framework.

Solves · Anticipating dynamic policy surrenders and their impact on asset-liability management (ALM).
  • Data pipeline & ETL feeding a Monte-Carlo simulation engine.
  • Predictive ML model with SHAP interpretability.
  • Stochastic rate calibration (Vasicek / Hull-White).
  • Dynamic surrender formula within a Solvency II framework.
  • FastAPI API & Streamlit dashboard; ALM financial-analysis report.

// stack

PythonMonte-CarloSHAPFastAPIStreamlitSolvency II

// 05. experience & education

The journey so far

Scroll through the path — each stop opens with its own photos.

  • ExperienceJune – Sep 2026

    AI & MLOps Intern (PFA)

    Vala Bleu — Web Host & IA · Agadir

    Photo 1
    • Building Vala Bleu Ops Copilot — an agentic RAG + MLOps assistant for IT Ops.
    • Multilingual FR/EN document RAG (pgvector) + server-log anomaly detection (Isolation Forest).
    • 100% local LLM inference on NVIDIA GPU; RAGAS evaluation, FastAPI / Streamlit, Docker.
  • ExperienceIn progress

    Research Intern (R&D) — Medical Prescription OCR

    École Nationale des Sciences Appliquées (ENSA) · Agadir

    Photo 1
    • Research work to extract structured medical information from handwritten Moroccan prescriptions, multilingual (FR/AR), evolving toward continual learning.
  • ExperienceJun – Aug 2025

    AI Developer — Intern

    AH Digital-Adopt IA · Technopark Agadir

    Photo 1
    • Built & deployed a production chatbot with MLOps monitoring (~60% faster response).
    • Improved RAG extraction accuracy from ~70% to 92%.
    • Delivered AutoExpert automotive platform (n8n + Supabase).
  • Education2022 – Present

    Engineering Degree — Big Data, Data Science & AI

    École Nationale des Sciences Appliquées (ENSA) · Agadir

    Photo 1
    • Cycle Ingénieur (Bac+5), specialty: Big Data, Data Science & AI.
    • Includes integrated preparatory classes.
  • Education2023

    Specialized Training — AI & Data Science

    Moroccan Digital Academy · UM6P

    Photo 1
    • Specialized program in Artificial Intelligence & Data Science.
  • Experience2024 – 2025

    President — Junior Enterprise JEEA

    ENSA Agadir

    Photo 1
    • Led a 7-person team; delivered 2 client projects.
    • Built 2+ company partnerships; official representation & negotiation.
  • Education2022

    Baccalauréat — Physical Sciences (First in Track)

    Houara High School

    • Graduated first in the Physical Sciences track.

// 06. certifications & recognition

Certifications & recognition

Recognitions, awards, and industry credentials.

Photo 1
Anthropic logo

Model Context Protocol: Advanced Topics

Anthropic

2026
Photo 1
Anthropic logo

Building with the Claude API

Anthropic

2026
Photo 1
Amazon Web Services logo

AWS

Amazon Web Services

2026
Photo 1
DataCamp logo

Data Engineer Associate

DataCamp

2026
Photo 1
HackerRank logo

SQL (Advanced) Certificate

HackerRank

2026
Photo 1
GITEX Africa 2024 logo

AI, Data Science & ML Masterclasses

GITEX Africa 2024

2024
Photo 1
IEEE / IFIP logo

IEEE_conference

IEEE / IFIP

2024
Photo 1
Taylor & Francis Group logo

Taylor & Francis — Research

Taylor & Francis Group

2026
Photo 1Photo 2
BCG X logo

Data Science & GenAI Virtual Experience

BCG X

2026
Photo 1
Deloitte logo

Data Analytics Virtual Simulation

Deloitte

2026
Photo 1
University of London logo

Professional Skills for International Business

University of London

2026
Photo 1
Make logo

Automation Certification

Make

2025
Photo 1
Bank of America logo

Investment Banking Job Simulation

Bank of America

2026

// 07. leadership

Not only a coder — I build teams too.

Leadership, community, and international programs, given equal weight to the engineering.

Photo 1
Leadership2024 – 2025

President

Junior Enterprise JEEA · ENSA Agadir

Strategic direction of a 7-member team — 2 client projects delivered, 3+ company partnerships, official representation and negotiation.

Photo 1
Leadership2023 – Present

Hackathons & Competitions

NASA Space Apps · engineering & AI challenges

NASA Space Apps Challenge Global Nominee; active participant in engineering and AI hackathons and competitions.

Photo 1
Leadership2024 – 2026

Events & Company / University Visits

ENSA Agadir · Oracle & partners

Organized and covered institutional events and company visits — including the official ENSA Agadir visit to Oracle.

Photo 1
Leadership2023 – 2024

Event Organizer & Host

Smart Home Startup · Injaz Al Maghrib

Communication strategy, visual identity, and digital marketing materials for a student startup.

Photo 1
Leadership2023 – Present

Clubs & Mentorship

Apps Club · ENSA Agadir

Technical workshops, peer mentorship and training, and hands-on development projects within the Apps Club.

Photo 1
Leadership2025 – Present

Event Moderator

Students' Association · ENSA Agadir

Moderating conferences and inter-school events — hosting panels, coordinating speakers, and running the stage.

Photo 1
Leadership2024 – 2025

Marketing & OGX

AIESEC Morocco & AIESEC Agadir

Marketing and coordination for international student-mobility programs.

Photo 1
Leadership2022 – Present

Sports & Cultural Events

ENSA Agadir

Participation in and organization of campus sports tournaments and cultural events.

Photo 1
Leadership2023 – Present

Social Activities

Student community · ENSA Agadir

Community initiatives, social outreach, and student-life activities across the school.

// 08. get_in_touch

Get in touch

Open to ML / AI / Data Engineering & research internships — and interesting conversations.

The fastest way to reach me is the terminal on the right, or any of the channels below. I reply to everything.

contact.sh

// let's work together

Let's build something
worth remembering.

I'm looking for ML / AI / Data Engineering and research internships where I can ship real systems and keep learning fast. If that sounds like your team, let's talk.

Open to final-year internships in ML / AI / Data Engineering & research