My Technology Watch in the AI Era

Foundations in PHP and Symfony Ecosystem

My technology watch primarily focuses on Symfony and more broadly on the PHP ecosystem, areas that form the foundation of my professional practice. As a developer, these technologies represent my core expertise and continue to evolve with constant innovations.

The Transformation of the Technological Landscape

It's evident that the IT world is currently undergoing an unprecedented transformation, catalyzed by the advent of artificial intelligence. This technological revolution is redefining not only our development practices but also the entire spectrum of business processes.

My Interest in AI and Automation

As part of my technology watch, I have a particular interest in two emerging fields:

Artificial Intelligence: Language models and generative AI systems now represent a revolution comparable to that of the internet.
Automation Processes: AI agents constitute the new technological eldorado, offering fascinating perspectives for streamlining and optimizing workflows.

The Challenge of Work-Study Programs in the Face of Technological Acceleration

As a work-study student, juggling between professional responsibilities and academic training represents a real challenge in keeping up with the exponential acceleration of technological evolution. However, this unique position offers me the opportunity to immediately apply theoretical knowledge in a practical context.

Exploration of Automation Tools

I'm particularly interested in N8N, an open-source process automation platform that stands out for its power and flexibility. This tool allows interconnection of different services and applications through an intuitive visual interface, facilitating the creation of complex workflows without requiring in-depth programming expertise.

Transition to Python for AI Development

To effectively keep up with AI advancements, I made the deliberate decision to learn Python, which has become the de facto language for AI and machine learning development. Despite my solid background in PHP, I recognized that Python's ecosystem offers unparalleled tools and libraries specifically designed for AI applications. This transition represents a significant but necessary learning curve in my professional development.

Venture into the Machine Learning Universe

With my growing Python skills, I started exploring DistilBERT, a compact language model (SLM - Small Language Model) based on the BERT architecture. Despite its interest, I quickly identified certain limitations inherent to this approach when working on more complex applications.

Democratization of LLMs in Local Environments

The HuggingFace platform, coupled with tools like Ollama, is revolutionizing access to AI technologies by enabling simplified installation and use of resource-efficient language models:

DeepSeek R1
Mistral 7B
And many other optimized models

This technological combination offers the possibility of leveraging LLMs (Large Language Models) locally, without exclusively relying on costly third-party APIs or cloud services.

The Strategic Issues of Digital Sovereignty

I remain particularly sensitive to issues of technical sovereignty and strategic governance, which constitute major challenges in the current context. Technological independence represents a fundamental pillar in ensuring control over our infrastructures and data.

The Need for Continuous Learning

Faced with the rapid emergence of these technologies, I realize the importance of continuous learning. I'm gradually trying to familiarize myself with new technical concepts that once seemed reserved for AI specialists:

JSON Squad: A format for structuring prompts and responses from language models

Inference: The process by which a trained AI model generates predictions or responses

RAG (Retrieval-Augmented Generation): An approach combining information retrieval and text generation that is revolutionizing how AI systems interact with data. By connecting language models to external knowledge sources, RAG enables more accurate and contextually relevant responses. In practice, this means AI systems can search through databases, documentation, or even the internet to supplement their responses with current and relevant information, rather than relying solely on their training data. I'm particularly intrigued by how RAG can be implemented to create context-aware applications that maintain up-to-date knowledge.

FAISS (Facebook AI Similarity Search): A powerful library developed by Facebook AI Research that enables efficient similarity searches and clustering of dense vectors. It's transforming how we interact with databases by allowing semantic search capabilities rather than just keyword matching. With FAISS, it becomes possible to search for concepts and meanings rather than exact terms, making information retrieval far more intuitive and powerful. I've begun experimenting with implementing vector databases using FAISS to create smarter search systems that understand the intent behind queries, not just the words used.

AI Agents: Perhaps the most fascinating development in recent AI progress is the emergence of AI agents—autonomous systems that can perceive their environment, make decisions, and take actions to achieve specific goals. Unlike traditional AI applications that perform discrete tasks, these agents can chain multiple capabilities together, reason about their progress, and adapt their approach based on outcomes. The potential applications are vast: from automated coding assistants that can build entire applications based on natural language descriptions, to research agents that can systematically explore scientific questions, to personal assistants that can seamlessly coordinate multiple services on our behalf. I'm particularly interested in the architecture of these systems, which often combines LLMs as the "brain" with specialized tools for different tasks, wrapped in a decision-making framework that can plan, execute, and evaluate actions in pursuit of objectives. While still in their early stages, AI agents represent a profound shift from passive to active AI systems that could fundamentally transform how we interact with technology.

These concepts, among others, constitute areas of exploration that I approach with curiosity and humility. I don't claim to be an expert in all these fields, but I strive to understand their fundamental principles and potential applications, especially as they relate to enhancing internet search capabilities and revolutionizing database interactions.

Conclusion: Being an Actor in the Digital Transition

The digital world is currently undergoing a major transition where traditional web, SEO, and many other areas are being profoundly disrupted by artificial intelligence. Faced with this revolution, I wish to position myself, at my level, as an engaged actor in this paradigm shift.

This active watch, although demanding, allows me to anticipate certain developments and gradually adapt my skills to the new realities of a constantly evolving sector. The learning path is long, but each new piece of knowledge acquired is a small victory in this technological race.