● The Advanced Institute for Artificial Intelligence (AI2) is an international consortium of leading researchers from diverse fields of Artificial Intelligence, dedicated to tackling socially relevant and high-impact challenges. ● This is a voluntary, non-remunerated research position focused on contributing to applied AI research and collaboration within the consortium.

Led the AI division at ActiveEon, driving innovation in Artificial Intelligence, MLOps, and large-scale automation. Specialized in orchestrating enterprise grade AI platforms to enable scalable, reproducible, and collaborative AI workflows across hybrid infrastructures (Cloud, HPC, and On-Premise). ● Led and mentored a team of 4 PhD-level AI researchers, driving innovation and cross-functional collaboration. ● Designed and scaled enterprise AI workflows, bridging research and production through automation and standardization. ● Architected distributed AI systems across multi-node and multi-GPU clusters, optimized for performance and cost efficiency. ● Built large-scale AutoML frameworks for hyperparameter optimization and neural architecture search (NAS). ● Drove AI industrialization and lifecycle automation, accelerating enterprise-wide adoption of AI-driven processes. Key Projects: ● Proactive AI Orchestration – Enterprise-grade AI orchestration suite integrating data pipelines, AutoML, MLOps, and collaboration tools. ● LLM & GenAI Integration – Developed MLOps extensions for large language and generative AI models, addressing scalability, observability, and infrastructure optimization. ● MLOps Dashboard – Real-time observability platform for AI/ML operations, including inference metrics, drift detection, and system-wide alerting. ● AI Catalogs & Services – Unified repository for reusable AI pipelines and ready-to-launch ML/Big Data services (e.g., H2O, TensorBoard, JupyterLab). ● AutoML & Optimization Engine – Scalable, multi-objective hyperparameter optimization framework supporting distributed tuning and neural architecture search. ● Python SDK & Jupyter Integration – Enabled seamless transition from notebook prototyping to production pipelines via automatic workflow conversion. Use Cases: ● Thales Alenia Space (TAS) – CNN-based anomaly detection on satellite hybrid circuit imagery. ● SAFRAN Aircraft Engines – AutoML-powered anomaly detection for aircraft engine sensor data. ● Hydro Québec (IREQ) – Machine learning and predictive analytics for electrical grid optimization. ● Desjardins Data Cup 2018 – Led feature engineering and model development for financial forecasting.

● Delivered a 10-hour graduate module focused on practical aspects of Machine Learning and Artificial Intelligence for Master’s students. ● The course covered hands-on ML/AI concepts, including supervised and unsupervised algorithms, Python-based implementations, and applied AI project development. ● Additionally, emphasized MLOps, AI in Industry, and AI Automation, exploring how to design, automate, and orchestrate end-to-end AI pipelines — from experimentation to production deployment. ● Topics included: workflow orchestration, CI/CD for ML systems, experiment tracking, model deployment, and pipeline automation. ● Promoted applied learning through industry-inspired case studies and reproducible AI practices.

Led the development and optimization of Large Language Models (LLMs) for specialized use-cases, focusing on fine-tuning open-source models like Llama2 7B and Mistral 7B. ● Designed AI solutions for real-world enterprise applications, focusing on LLM fine-tuning, model optimization, and efficient AI deployments. ● Provided AI consulting for customers to integrate and optimize AI workflows. ● Developed end-to-end LLM pipelines covering data collection, fine-tuning, evaluation, and model serving. ● Implemented parameter-efficient fine-tuning (PEFT) strategies using LoRA. ● Optimized inference performance with FlashAttention-2 and AutoAWQ quantization. ● Contributed to the open-source miniogre project.

CamPark Solutions is a French start-up company focused on computer vision solutions for smart cities. ● Worked as a Computer Vision and Deep Learning consultant for the automated parking lot classification project. ● Developed a C++ algorithm based on deep convolutional neural networks for image classification, optimized to run on a Raspberry Pi board.

GeekSys is a Brazilian start-up company focused on computer vision and big data analytics. ● Worked as a Computer Vision consultant for the Heatmaper project.

● Developed enterprise Java-based applications and mobile apps using HTML5 and Sencha framework.

● Designed and developed J2EE-based enterprise applications for the healthcare sector.

● Developed a vehicle fleet management system integrating J2EE and SAP.

● Led software architecture for the SEIA project, an environmental information system. ● Defined system design constraints, researched SOA technologies, and developed software components.

● Designed and developed web and desktop applications for telecommunications, specializing in Java, PHP, J2EE, and database systems (SQL, MySQL, PostgreSQL).