I'm a Full Stack Engineer and AI enthusiast currently pursuing an Engineering Degree in Computer Science & Data Science at École des Mines de Saint-Étienne (France).
I specialize in building end-to-end systems that merge modern web technologies with intelligent automation, leveraging AI models for perception, reasoning, and user experience enhancement.

My work spans from scalable web platforms to experimental AI agents that operate visually and autonomously.

A fully autonomous Reddit agent that interacts with posts and comments based on natural language prompts.

• Built a visual AI agent that reads, interprets, and acts on Reddit posts using screenshots only — no HTML or API calls.
• Users can input prompts like “Boost my karma by helping others”, and the agent autonomously comments, upvotes, or downvotes accordingly.
• Utilizes a fine-tuned GroundingDINO model trained on a custom dataset of Reddit UI scraped via Puppeteer.
• Can run 100% locally (16GB VRAM) or in a hybrid mode using Perplexity API for NLP.
• Demonstrates multi-modal reasoning, reinforcement-like behavior, and real-world AI-human interaction.

Tech Stack: PyTorch, GroundingDINO, Perplexity API, Puppeteer, Python, Computer Vision, LLMs

Note: I scraped Reddit to generate screenshots and annotations in COCO format using RedditScraper and used my personal powerful GPU (RTX 5080) to fine-tune GroundingDino entirely locally.

A cross-border marketplace connecting buyers and proxies for rare or affordable products worldwide.

• Founded and developed from scratch: from architecture to deployment.
• Full-stack solution using Next.js 15, Spring Boot, and AWS with Stripe and Wise integration.
• Real-time communication with WebSockets and a responsive, elegant UI built with TypeScript and TailwindCSS.
• Built to scale and handle international transactions with secure payment and identity layers.

Tech Stack: Next.js, Spring Boot, TypeScript, AWS, Stripe API, Wise API, TailwindCSS, Docker

• Vision-Language Models (VLMs) and multimodal AI
• Data pipelines for intelligent agents
• Reinforcement learning for web-based agents
• Advanced TypeScript patterns and Next.js 15 features

• Arabic: Native
• English: C2 (TOEIC 930)
• French: C1

• Email: [email protected]
• GitHub: Marabii
• LinkedIn: linkedin.com/in/dadda-hamza-997616324

• Build intelligent agents that interact naturally with digital environments.
• Continue developing Jeebware into a scalable AI-powered platform.
• Find an interesting internship position where I can demonstrate my skills in both AI and web dev.

— Hamza DADDA