Senior AI Engineer

I own end-to-end delivery of production AI Employees for venture capital and clean-energy clients, from requirements and statements of work through technical design, staging, and production launches. I build conversational and agentic systems with multi-turn state, per-user memory, streamed tool execution, web chat, and Telegram/iMessage access.

I design secure MCP tool layers, pgvector RAG assistants, structured LLM workflows, and integrations with Affinity CRM and Microsoft 365 Graph. I also develop FastAPI and NestJS services, Supabase/Postgres data models, React/TypeScript interfaces, JWT/RBAC security, and maintain production quality through 1,580+ automated tests.

I designed agentic LLM flows for candidate qualification, matching, and resume generation using OpenAI, Gemini, Claude, and Llama models. I implemented prompt engineering, embeddings, caching, strict output validation, Langfuse observability, and feedback loops for iterative matching-quality improvement.

I led development of a hiring intelligence platform with FastAPI microservices, OCR document-processing pipelines, and hybrid Elasticsearch/vector search. I also architected AWS EMR Spark ETL processing over 800 million profiles, deployed event-driven SQS pipelines, and delivered Dockerized ECS services using Terraform and monitoring alerts.

I engineered semantic talent-matching pipelines using BERT, Doc2Vec, GloVe, cosine similarity, RankNet learning-to-rank, and human feedback. I built modular LangChain architectures that supported local Ollama and cloud-based Groq and Hugging Face inference.

I developed NLP, OCR, document digitization, classification, and computer-vision solutions using Python, TensorFlow, Flask, PaddleOCR, and Scikit-learn. I deployed services through Docker and AWS infrastructure provisioned with Terraform, with CI/CD automation via GitHub Actions.

I trained more than 20 computer-vision models for microorganism detection and counting, achieving over 85% mean average precision. I improved model robustness using data augmentation and translated stakeholder requirements into model and pipeline designs.

I developed Python, PyTorch, FastAPI, and Uvicorn APIs and implemented MLOps architecture with Airflow, MLflow, and DVC. I deployed containerized systems on AWS, GCP, and DigitalOcean, automated delivery through Bitbucket Pipelines, and reduced GPU training costs by 65% using AWS spot instances.

I contributed to RLHF pipelines by generating high-quality training examples and evaluating ChatGPT responses for Python code generation. I validated and ranked model outputs for quality, relevance, and correctness.

I developed optimized Python solutions for complex tasks, documented reasoning and implementation details in Jupyter notebooks, and participated in code reviews, debugging, and technical documentation.

I designed and deployed deep-learning models for tomato plant disease detection, preparing and labeling more than 3,000 images. I evaluated YOLOv5, YOLOv7, and YOLOX models, with YOLOv5 reaching 92% mAP.

I improved detection accuracy to 93.1% through hyperparameter evolution and deployed the final model in an Android application. I used Python, Keras, TensorFlow Lite, Google Colab, and Android Studio.

I led IT operations, service management, strategic projects, and cross-functional collaboration across facilities, HR, and project teams. I migrated legacy infrastructure from ten physical servers to a virtualized VMware cluster and administered Oracle, MySQL, and PostgreSQL databases.

I developed ETL pipelines from MySQL, Oracle, Excel, and CSV sources, delivered BI dashboards, and built a real-estate apartment price-prediction model to support business decision-making.

I implemented the OPTIMA project to improve telephone reception efficiency for a major French customer. I defined functional and technical architecture for multiple modules and ensured the design aligned with business requirements.

I produced installation documentation and user guides, supported deployments, participated in customer conferences, and helped negotiate feature enhancements based on client needs.

I developed configuration tools and software components to automate telecom customer site pre-configuration using engineering rules. I generated installation documentation in PDF, Excel, text, and SVG formats.

I coordinated with commercial, logistics, and technical teams and provided frontline customer support to resolve issues and maintain delivery continuity.

I volunteered as a Cloud Engineer supporting scalable AWS infrastructure for business applications. I configured S3 buckets with strict access controls and provisioned EC2 spot instances for cost-efficient deep-learning training.

I also implemented granular IAM roles and policies to enforce least-privilege access, security, and operational reliability.

Master's degree in Data Engineering completed in Casablanca, Morocco.

Software Engineering degree completed in Rabat, Morocco.

Preparatory classes in MPSI.