Omar Nouih | AI & Data Engineer

Professional Journey

Work Experience

Chef de Projet Data & IA

DataScale Business

Full-time Jan 2026 – Present

Clients Managed

Teams Coordinated

100%

On-time Delivery

Leading end-to-end data & AI project delivery at DataScale Business — managing technical teams, driving innovation in BI and AI solutions, and acting as primary technical liaison for clients across multiple industries including retail, finance, and logistics.

Key Responsibilities

Piloting AI product roadmaps from scoping and architecture through deployment, ensuring alignment between business objectives and technical feasibility at every milestone.
Coordinating cross-functional teams (data engineers, ML engineers, BI developers) to deliver custom AI and BI solutions: predictive models, NLP engines, and executive dashboards tailored per client.
Establishing data governance frameworks and best practices for clients migrating to cloud-based data architectures, covering access controls, lineage, and quality checks.
Translating complex business requirements into actionable data engineering tasks, sprint backlogs, and deliverables — serving as the bridge between stakeholders and engineering teams.
Running workshops and technical demos for C-level clients, presenting AI capabilities, BI insights, and ROI projections to drive buy-in and expansion.

Tech Stack

LLMsPower BIPythonETL/ELTCloud BILooker StudioProject ManagementAzure

Data Engineer

DataScale Business

Full-time Nov 2025 – Jan 2026

10+

Pipelines Built

Dashboards Delivered

AI Solutions Deployed

Designed and built on-premise and cloud BI architectures, engineered robust data pipelines, and delivered AI-powered analytical solutions tailored to clients' operational needs across diverse industries.

Key Responsibilities

Architected ETL/ELT pipelines connecting heterogeneous data sources (ERP, CRM, flat files, APIs) into unified analytics layers on both on-premise SQL Server and cloud environments.
Developed custom AI solutions including predictive demand models, NLP text classification engines, and conversational business assistants using Python and LLM frameworks.
Built interactive Power BI and Looker Studio dashboards with real-time data refresh, drill-through analytics, and KPI monitoring for client executive teams.
Implemented data quality validation layers — automated anomaly detection, schema drift alerting, and reconciliation reports to ensure pipeline reliability at scale.
Collaborated directly with client data teams to document requirements, design data models (star/snowflake schemas), and optimize SQL query performance for reporting workloads.

Tech Stack

PythonETL/ELTPower BILooker StudioSQL ServerPostgreSQLNLPPredictive ModelsAzure

Machine Learning Engineer

Labelvie Group

Freelance Sept 2025 · 1 month

10K+

Products Analyzed

CamemBERT

Model Used

XGBoost

Best Regressor

As a Freelance ML Engineer, analyzed e-commerce product datasets (10K+ products) and developed an automated pipeline to estimate product weight attributes directly from textual labels — improving data quality and supporting business analytics initiatives at Labelvie.

Key Achievements

Analyzed and cleaned a dataset of 10,000+ e-commerce product entries, handling missing values, outlier detection, and label normalization for product weight attributes.
Trained CamemBERT-based NLP models with transfer learning to extract weight signals from French product names and descriptions — leveraging pre-trained bilingual embeddings.
Combined NLP embeddings with regression models (XGBoost, Ridge, Linear Regression) for robust weight predictions across diverse product categories (food, hygiene, electronics).
Built a complete preprocessing and feature engineering pipeline: unit extraction via regex, numeric pattern recognition, and category-aware normalization for 10K+ SKUs.
Delivered the project fully remote, providing well-documented Jupyter notebooks, model artifacts, and a final report with performance benchmarks across product subcategories.

Tech Stack

CamemBERTTransfer LearningXGBoostNLPPythonScikit-learnPandasRegex

Data Scientist — PFE Intern

Labelvie Group

Internship (PFE) Mar 2025 – Aug 2025

95%

SQL Accuracy

80%

Reporting Time Saved

1K+

Daily Queries Tested

Master's thesis internship: designed and developed a decision-support assistant enabling business teams to query retail data using natural language (French/Arabic), deployed directly on WhatsApp. Also automated Excel reporting and built real-time BI dashboards.

Key Achievements

Designed a natural language to SQL chatbot (GPT-4o + LangChain) that interprets complex business questions in French/Arabic and generates accurate, sanitized SQL against Azure SQL retail databases.
Deployed the assistant on WhatsApp via Flask + WhatsApp Business API with role-based access control (RBAC), query sanitization layers, and full audit logging for data governance compliance.
Implemented a data quality monitoring system with Python validators, schema drift alerts, and performance diagnostics for high-volume queries on confidential retail datasets.
Automated Excel business reporting workflows with Python (openpyxl, pandas) — eliminating manual copy-paste processes and saving the team ~80% of weekly reporting time.
Designed Power BI and Looker Studio dashboards with real-time data refresh, enabling non-technical managers to self-serve sales KPIs, stock levels, and supply chain metrics.
Containerized the full solution with Docker for reproducible deployment across dev, staging, and production environments.

Tech Stack

GPT-4oLangChainAzure SQLFlaskWhatsApp APIPower BILooker StudioDockerPythonNLP

Data Scientist & Lead Developer

3D Smart Factory

Internship (PFA) Jul 2024 – Sep 2024

87%

Match Accuracy

60%

Screening Time Saved

400+

CVs Processed

Developed RucRut — an AI-powered recruitment platform automating CV analysis, semantic candidate-to-job matching, and LLM-generated interview questions. Built end-to-end from NLP models to a full-stack RESTful application with an analytics dashboard.

Key Achievements

Built an automated CV parsing engine extracting structured skills, experience, education, and contact data from unstructured PDF/Word resumes using Hugging Face Transformers.
Implemented vector similarity search with Sentence Transformers (all-MiniLM-L6-v2), achieving 87% candidate-job matching accuracy — a major leap over keyword-based filtering.
Designed LLM-powered interview question generation (LLaMA-based) that dynamically tailors question sets per candidate profile and job description — covering technical, behavioral, and situational questions.
Built a scalable Flask + MongoDB RESTful backend with JWT authentication, paginated endpoints, and a modular pipeline architecture supporting concurrent candidate evaluations.
Developed an Angular analytics dashboard visualizing recruitment funnel metrics, candidate scoring breakdowns, and skill gap heatmaps for hiring managers.
Reduced manual CV screening time by 60%, enabling recruiters to focus on high-value candidate interactions instead of administrative filtering.

Tech Stack

Hugging FaceSentence TransformersLLaMAFlaskMongoDBAngularPythonScikit-learnNLTK

Chatbot Developer

Royal Air Maroc

Internship (PFE) Apr 2023 – Jun 2023

92%

Intent Accuracy

100+

Active Users

75%

HR Tickets Reduced

Developed and deployed an intelligent HR virtual assistant for Royal Air Maroc, automating employee requests for leave balances, payslips, and work certificates — serving 100+ staff members directly on WhatsApp.

Key Achievements

Built a Rasa NLU chatbot with custom SpaCy NER models covering 50+ distinct HR intent scenarios, achieving 92% intent classification accuracy on held-out test sets.
Integrated the chatbot with legacy HR systems via secured Flask REST APIs — enabling real-time retrieval of leave balances, payslip PDFs, and dynamic work certificate generation.
Deployed to WhatsApp Business via Twilio, providing 24/7 automated HR support and reducing manual HR ticket volume by 75% within the first month post-launch.
Implemented multi-language support (French and Arabic) within the dialogue manager, allowing seamless switching based on user language preference during conversations.
Built a conversation analytics dashboard (Python + Matplotlib) tracking intent distribution, user satisfaction scores, and knowledge gap identification — fueling continuous NLU model improvement.
Used Redis for session state management, ensuring fast context retrieval across multi-turn conversations for 100+ concurrent employee sessions.

Tech Stack

Rasa NLUSpaCyFlaskTwilioWhatsApp APIPostgreSQLPythonRedis

What I Built

Featured Projects

Top Project

RucRut — AI Recruitment Platform

An intelligent recruitment platform that automates the entire hiring pipeline. It parses CVs automatically, semantically matches candidates to job requirements, and generates tailored interview questions using LLMs. Built end-to-end as PFA intern at 3D Smart Factory.

Hugging FaceSentence TransformersFlaskMongoDBAngularLLMsPythonScikit-learn

View on GitHub Demo

Automated CV Parsing

Extracts structured data from PDF and Word resumes using Transformer-based NLP — turning unstructured text into actionable candidate profiles.

Semantic Matching Engine

Vector similarity search with Sentence Transformers captures nuanced skill and experience alignment, far beyond keyword matching.

LLM Interview Generator

Generates custom interview questions tailored to each candidate profile and job description using large language models.

87%

Match Accuracy

60%

Screening Saved

400+

CVs Processed

NL to SQL Chatbot (Labelvie)

Enterprise conversational analytics assistant (GPT-4o + LangChain) deployed on WhatsApp. Translates natural language business questions into precise SQL on Azure databases with RBAC and audit logging.

GPT-4oLangChainAzure SQLFlask

HR Chatbot (Royal Air Maroc)

Rasa NLU + SpaCy HR assistant handling leaves, payslips and certificates for 100+ employees. Deployed on WhatsApp via Twilio. 92% accuracy, 75% ticket reduction.

RasaSpaCyTwilioFlask

Real-Time Twitter Sentiment Analysis

End-to-end NLP streaming pipeline using Apache Kafka and Spark Streaming. Real-time sentiment classification with a live web dashboard visualizing social media trends.

KafkaSpark StreamingNLPPython

NYC Taxi Data Pipeline

Production-grade ETL pipeline with Airflow orchestration, Kafka ingestion, Spark processing, dbt transformations and a PostgreSQL analytics store.

AirflowKafkaSparkdbtPostgreSQL

FSTT Academic Chatbot

University assistant powered by RAG, LangChain, and vector databases. Scrapes and indexes FSTT content for intelligent student Q&A sessions.

LangChainRAGVector DBLLM

VersaLearn — Data Analysis Tool

Interactive desktop Python tool for data analysis and ML experiments with a Tkinter UI, Pandas, Scikit-learn model training, and Matplotlib visualizations.

PythonTkinterScikit-learnMatplotlib

View All on GitHub