Saida, Saïda, Algeria
I’m AbdElKader Seif El Islem RAHMANI, a Deep Learning Engineer and PhD Researcher in Artificial Intelligence passionate about building self-improving, autonomous, and generative AI systems that move beyond theory into real-world impact. At Systelium (France), I help organizations leverage AI-driven digital transformation—from intelligent document recognition and report generation to autonomous agent integration. My work sits at the intersection of Deep Learning, NLP, and Generative AI, turning research breakthroughs into scalable, production-ready systems. Currently pursuing my PhD at the University Center of Naâma, I explore Generative AI methods for autonomous agents and self-improvement behavior. My mission is to design AI that not only solves complex problems but also learns how to learn—optimizing itself over time. I’ve led and contributed to award-winning projects, including AgriChat (RAG + Mixture of Experts), which boosted advisory efficiency by 35%, and cloud-based fingerprint recognition, improving accuracy and latency for secure enterprise use. My technical toolkit includes Python, TensorFlow, PyTorch, Keras, Hugging Face, Docker, FastAPI, and CI/CD pipelines, combined with a deep understanding of LLMs, RAG, MoE, RL, and Bandit Algorithms. I’m driven by curiosity, collaboration, and the belief that AI should evolve as intelligently as the problems it solves. 📩 Always open to collaborations in AI research, product innovation, or applied deep learning—let’s connect and build the next generation of intelligent systems.
Led end-to-end development and deployment of an enterprise construction management platform as sole technical lead across 70+ production releases. Built a secure, scalable backend using Python (FastAPI), async architecture, and PostgreSQL; implemented CI/CD pipelines reducing deployment cycles by 60% while maintaining 99.9% uptime. Integrated AI automation using Google Gemini LLMs for document classification, data extraction (CCTP, CCAP, PPSPS), automated task generation, and natural-language workflows—reducing manual processing by 70% and enabling real-time decision-making. Architected 50+ RESTful microservices covering project management, AI-driven planning, RBAC, stock/attendance tracking, alert systems, and organizational structures, with async optimization and robust error handling. Implemented advanced business logic including retro-planning with AI delay prediction, automated validation, multi-tier alerts, supplier payment tracking, S3 document versioning, and real-time dashboards—improving project completion rates by 40%. Managed database evolution with 100+ Alembic migrations, ensuring idempotency and zero-downtime deployments. Built production infrastructure with Docker, Nginx, SSL/TLS, multi-environment orchestration, monitoring/logging (APScheduler), and JWT-based security. Delivered high-quality engineering through testing (pytest async), documentation, and cross-functional collaboration. Led backend architecture, AI workflows, and platform integration, including authentication, CRUD APIs, file handling, and async data models. Developed AI features such as project-aware chat and document processing pipelines, and established monorepo practices for scalable development. Key Metrics: 6k+ commits | 70+ production PRs | 10+ microservices | 50+ APIs | 100+ Succeded automatic deployment | Full-stack ownership from data to deployment.
Maintained open-source ML libraries improving data preprocessing efficiency and model reproducibility for 1 000 + global users. Published reusable codebases integrating LLMs, CV, and RL algorithms — adopted by academic and industry collaborators. Reviewed pull requests, enhanced documentation, and promoted clean coding and modular design, ensuring project scalability. Optimized key workflows, reducing computational cost by up to 30% through algorithmic and data-pipeline improvements.
Designed and deployed end-to-end deep learning solutions in computer vision, NLP, and generative AI with TensorFlow, Keras, and PyTorch. AgriTechly: Improved plant disease detection accuracy by 15% and reduced inference time by 20%, earning 2nd place in a national AI hackathon. Fingerprint Recognition in Cloud: Delivered a 29% accuracy gain and reduced latency by 300 ms, enabling secure and fast biometric authentication. AgriChat (LLM Expert System): Increased advisory efficiency by 35% using RAG + Mixture of Experts (MoE) for real-time agricultural support. Delivered full-stack ML deployments with Flask, FastAPI, and Docker, ensuring maintainable pipelines and reproducible experiments.
Authored and shared top-ranked Kaggle notebooks, helping peers replicate high-performing ML pipelines in vision and NLP tasks. Developed reproducible solutions emphasizing accuracy, efficiency, and explainability, increasing baseline performance by 10–25% across multiple datasets. Contributed to community discussions and model reviews, improving collective project outcomes and expanding open data resources.
Delivered 25 + custom software projects (web, mobile, and desktop) using React, Node.js, and Python, achieving 100% client satisfaction. Built secure backends with PHP (Laravel) and Python APIs, integrating RESTful services and optimizing query performance by up to 50%. Implemented AI-powered features (chatbots, recommendations) to enhance UX and automation. Collaborated with clients to design scalable architectures and delivered all projects on time and within budget.