About
I’m the founder and CTO of Fernus, building AI-driven educational technology systems.
Full-stack developer, designer, and an old-school engineer who started with assembly, CGI, Perl, and the early days of PHP. I’ve been a PHP developer since its early years, but I’ve never been tied to a single language or platform.
I don’t choose technologies. I choose problems.
One of the advantages of being an old-school developer is adaptability. I can pick up new languages and tools quickly, because the fundamentals don’t change. Systems, constraints, and architecture always come first.
My journey started around the age of 12 with machines like the Atari 800XL, Commodore 64, and Amiga. That background shaped how I approach technology today. I don’t just write code, I build systems.
Over the years, I’ve developed platforms that reached Alexa’s top 10, and since 2007, my main focus has been educational technology. In recent years, I’ve been working deeply on AI systems, not as experiments, but as production-grade solutions.
My work in AI focuses on practical, system-level implementations. I build pipelines for educational content processing, including classification models, embedding-based retrieval systems, and encoder-decoder architectures. These systems are designed to analyze questions, map them to learning outcomes, estimate difficulty levels, and generate structured insights.
I work extensively with embeddings and semantic representations to enable similarity search, clustering, and knowledge mapping. Retrieval-augmented generation (RAG) plays a key role in combining structured educational content with generative models.
I’m particularly interested in how AI can be used to understand and generate educational content at scale. This includes question generation, solution analysis, content tagging, and automated evaluation systems. The focus is always on accuracy, explainability, and real-world usability.
I’m not interested in AI as a trend. I’m interested in AI as infrastructure.
At Fernus, I design and build systems that combine AI, data, and real-world constraints. The goal is clear. Scalable, reliable, and useful systems that actually work in production.
I’ve always preferred going deeper instead of wider. Whether it’s modifying retro hardware, designing microcontroller-based circuits, or optimizing low-level systems, I like understanding the full stack.
Outside of work, I spend time with retro computing, DIY electronics, and hands-on projects. Music, gardening, and motorcycles help keep things balanced. I also play in the Ankara PDO Dart League.
Some people know me as “Fernus Serkan”. That works.
What You’ll Find on This Blog
- Educational content processing and AI pipelines
- Classification models, embeddings, and semantic systems
- Encoder-decoder architectures and generative workflows
- RAG systems and scalable AI infrastructure
- Microcontrollers, minimal systems, and hardware projects
- Real-world engineering decisions
No fluff. Just systems that work.