[nom]

[e-mail] ⋄ Portfolio ⋄ LinkedIn ⋄ HuggingFace ⋄ Github

PROFILE

AI Engineer specialized in speech processing and NLP for low-resource languages, with experience build-ing AI-based SaaS applications, voice analytics platforms, and voice agents. Skilled at delivering end-to-end, cloud-ready ML pipelines, from data ingestion to production, that combine speech-to-text, data privacy/PII anonymization, and software engineering for multilingual environments, with deep focus on African languages, primarily in the customer experience domain.

EDUCATION

Engineering Cycle, Artificial Intelligence, Euromed University of Fes · 2021 - 2024

Preparatory Classes, Euromed University of Fes · 2019 - 2021

Baccalaureate, Experimental Sciences, With High Honors · 2019

EXPERIENCE

AI Engineer at ToumAI — Dec 2024 - Present

Key Responsibilities

• Building speech-to-text models for multilingual dialects, namely Moroccan and Tunisian Darija.
• Building on-premise data-privacy pipelines for voice and PII anonymization.
• Building reproducible ML workflows from data ingestion to prediction generation.
• Automating & integrating AI components as micro-services and event-driven workflows, connecting with orchestration tools and business APIs to automate decision pipelines.
• Analyzing clients’ business requirements, drafting analytics KPIs, and translating them into technical data-science problems.
• Collaborating with business, product, and engineering teams to build end-to-end data-science pipelines.

Key Achievements

• Delivered 2 speech-to-text models for Moroccan and Tunisian Darija, used in both the audio data- privacy and analytics layers, as fine-tuned versions of different architectures (e.g. OpenAI Whisper, Qwen- ASR, NVIDIA NeMo).
• Delivered 2 Named Entity Recognition models for Darija and Tunisian PII detection, built on archi- tectures such as BERT-based models, GLiNER, etc.
• Delivered the analytics layer of the startup product HolisticX ([lien]), capable of analyzing both enterprise phone call centers and virtual voice agents to extract business insights, agent performance trends, and customer preferences using speech-to-text, audio diarization, and emotion detection; rolled out for many leading Moroccan telecom call centers.
• Delivered a Microsoft-certified SaaS solution (Microsoft Marketplace) that enables integrating Multilin- gual AI Assistants with the Microsoft Dynamics 365 platform.

Data Scientist at DeepEcho — Feb 2024 - Nov 2024

• Conducted research on adapting Vision-Language Models (e.g. CLIP, BLIP-2, LLaVa-Med) to fetal ultrasound data through parameter-efficient fine-tuning (e.g. LoRA), involving data curation, finetun- ing, and model evaluation using BLEU, ROUGE, BERT-SCORE and Clinical F1-score for ultrasound image-text understanding.
• Implemented an image classification pipeline to categorize fetal ultrasound images by type (e.g. 3D ultrasound, Doppler, standard 2D scans, etc.)

Internship at L’ODJ Média — July 2023 - Aug 2023

• Developed an All-in-One automated media content generation platform, integrating news scraping, LLMs for article generation, TTS models for voice synthesis, and diffusion models for image/video generation with human-in-the-loop oversight, enabling end-to-end content production and automated streaming to the company’s news channels.

Internship at SAFRAN Aerosystems — July 2022 - Oct 2022

• Implemented a full-stack web application with Laravel, featuring Front-End components, RESTful Back-End services, and a MySQL database to digitize and manage aircraft component documentation and reporting workflows.

SKILLS

Programming Languages · Python, C++, C. Familiar with Java

Frameworks & Libraries · Pytorch, Transformers, Unsloth, vLLM, LangChain, LangGraph,

Pipecat, FastAPI

AI Techniques · RAG, MCP, Agentic Workflows & Tool Use, Prompt Engineering,

Efficient Fine-tuning, Model Quantization, vLLM Deployment

Embeddings & Vector Search, Speech-to-Text, LLM Evaluation

Database Management · Vector Databases (QDrant), ClickHouse, MongoDB

Version Control & Cloud Tech · Git, GitHub, Docker, Azure AI Foudry, Vertex AI, AWS Bedrock

VOLUNTEER

Contributor at AtlasIA: A community-driven collective building open-source Moroccan datasets and AI mod-els for Darija; contributed speech-to-text, vision-language, and NLP datasets and models for underrepresented

North African dialects.

Mentor at HackAI - UM6P - 1337AI: Collaborated on the development of educational materials and mentored participants during the HackAI Morocco 2025 and 2026 hackathons, a national event focused on Generative AI.

Mentor at Teens in AI - ICESCO-UEMF Women in Science: Mentored youth during the Teens in AI hackathon, guiding teams on AI concepts and demonstrating how to apply AI technologies to bring their project ideas to life.

LANGUAGES

Arabic: Native · English: Professional French: Professional Spanish: Beginner