I work on large-scale distributed AI systems, modeling LLM training and inference workloads as execution traces and simulation scenarios that drive cluster capacity, hardware selection, and network topology decisions.
Mustafa Mert Tunalı
I build machine learning systems that run in production. These days, I work on distributed AI simulation at Turba Labs, modeling how training and inference workloads behave at scale.
Before that, I worked on document understanding, computer vision, and generative AI. Away from the keyboard: racing, science fiction, and neuroscience.
Work
May 2026–present
Previously, at Multimodal
At Multimodal, I designed and built the architecture behind AgentFlow and Instant, taking both from initial design to production.
Open source
go-chatgpt-grpc
Go / gRPCA Go server for streaming and custom backend features around the OpenAI API.
Deep Learning Training GUI
Python / MLA desktop interface for training pretrained models and following progress in TensorBoard.
Research
My published work is in computer vision for industrial quality control. I’m also interested in ML infrastructure and bio-inspired AI.
Off work
You’ll find me at the Nürburgring Nordschleife or in a racing sim, trying to improve every apex and braking point.
I read science fiction and neuroscience papers when I can. The connection between biological and artificial intelligence keeps me curious.
Nürburgring Nordschleife
Education
MEF University
2020–2024BSc in Computer Engineering