Dağlar Berk Erdem’s route into software did not begin in a computer science department. It crossed chemical engineering, industrial process work, computer-vision research, and the practical systems around machine learning.
Today, he is a Platform Engineer at FAL.ai, the generative media platform. His short submitted biography keeps the description simple. He works on infrastructure. His public record provides a fuller picture of the engineering path behind that sentence.
More than one kind of engineering
Dağlar studied Chemical Engineering at Boğaziçi University and began his career as a process engineer. He later moved into software and graduate research in the university’s Computer Engineering department.
That change of field is less abrupt than it first appears. Both kinds of work require somebody to understand a system as a set of connected constraints. The materials differ, but measurement, reliability, and trade-offs still matter. This is our editorial reading of his path, not a claim Dağlar has made about himself.
His graduate work focused on computer vision, deep learning, and gait recognition across different camera viewpoints. In 2024, he presented research on viewpoint dependency in gait recognition at the IEEE Signal Processing and Communications Applications Conference.
The problem is easy to state even if the models are not: the same movement can look different when a camera sees it from another angle. Studying that gap means testing what a model has actually learned instead of relying on a single favorable view.
Making model operations concrete
Dağlar’s public repositories show the operational side of that background. His MLOps project works through repeatable model execution with ClearML, containers, infrastructure as code, worker management, and monitoring.
It is not a glossy product page. It reads like an engineer’s working notebook, with setup steps, scripts, dependencies, and the details needed to reproduce the environment. That makes it useful evidence of how he approaches systems, without requiring a list of broad claims.
Beyond the keyboard
There is another long-running discipline in Dağlar’s public history. He competed in swimming and water polo, and the Kınalıada Water Sports Club archive lists him in its senior water-polo squad for the 2016 to 2017 season.
We will not turn that into a tidy lesson about engineering. It is simply part of getting to know the person behind the job title.
We are very happy to welcome Dağlar to the first LatentShift edition in İstanbul on 17 October. Before then, you can explore his work on GitHub or connect with him on LinkedIn.