From Signals to Intelligence

I started as an electronics engineer, working close to signals rather than data science. My graduation project, RASED, combined radar, signal processing, and embedded systems.

That naturally led me to the ITI Wireless Communications track, where I expanded into wireless systems, embedded development, DSP, image processing, visible-light communication, and localization.

In February 2017, I joined Orange as an Access Transmission Planning Engineer.

The role taught me how to coordinate requests, lead work across teams, and understand how real telecom networks are operated.

But something felt incomplete. The work was heavily manual, automation was still rare across the industry, and much of my technical knowledge was being used only to execute recurring actions.

I was learning how organizations operate — but I wanted to understand and build more.

Knowledge was fragmented. To understand one problem deeply, I often had to find the right person, ask the right question, and reconstruct the complete picture myself.

Knowledge everywhere. Complete picture nowhere.

I spent nearly six months asking a different question: if this wasn't the path I wanted for the next decade, what was?

In 2017, AI was far from an obvious career choice. The field was difficult, uncertain, and many engineers around me were unsure whether it would become a sustainable profession.

After speaking with people inside and outside the industry, I reached a conclusion: the risk was real — but so was the opportunity.

The future would need engineers who understood the domain, the mathematics, and the programming well enough to see problems others could not.

I decided to make that bet.

I joined Nile University with partial funding and deliberately chose AI rather than specializing further in one telecom subsystem.

Telecom specialization could have made me deeper in one part of the network. AI could give me a capability that could move across the entire network.

The first real proof came while I was still working in transmission planning.

I initiated and directed the development of an IP capacity dimensioning and offline radio-site configuration tool connected directly to network systems.

It automated work that previously consumed roughly a full working day and combined multiple checks into one workflow.

It wasn't just faster reporting. The tool reconstructed enough network context to identify problems that previously required multiple manual checks.

The tool worked. People used it. My name started travelling further than my job title.

When the Telecom Cloud manager approached me, he gave me a different proposition.

I could move directly toward AI — or spend two demanding years learning the systems around it: infrastructure, deployment, operations, and production engineering.

I chose the harder option.

Those two years filled a gap I had not fully understood yet.

AI taught me how to build models. MLOps taught me how to make them survive reality.

While developing production and MLOps skills, I was also completing my master's and building a research track in parallel.

That period resulted in five publications and gradually changed how I viewed industry and academia: they did not have to be separate careers.

Industry gave me difficult real problems. Research gave me better ways to formulate them.

When I started the master's, my assumption was simple: learn AI well, then leave telecom.

By the time I finished, I had reached almost the opposite conclusion.

Telecom wasn't something I needed to escape. It was the domain advantage that made my AI skills different.

The question was no longer “AI or telecom?” It became “What can I do because I understand both?”

My next move had three goals: broader career exposure, experience in more advanced telecom environments including 5G, and a role where telecom, data science, and AI were no longer separate tracks.

In my current environment, the pieces finally converged.

For the first time, I wasn't choosing between the things I had learned. I was using all of them.

The next decade is not about adding another tool to the stack. I want to work on problems difficult enough to influence how future intelligent networks are designed, evaluated, and operated.

That means moving deeper into research: working in strong research environments, contributing to foundational work, and eventually helping shape the methods and standards others build upon.

The first deliberate step is the PhD.