Hi! I'm Meltem.
About. I'm a PhD student in Building Performance & Diagnostics at Carnegie Mellon University, advised by Azadeh Sawyer and Vivian Loftness. My research sits at the intersection of multimodal AI, spatial reasoning, and building science. I develop methods to understand existing building stocks from sparse, heterogeneous, and incomplete data, with applications in urban-scale energy modeling, building simulation, retrofit optimization, and grid impact analysis.
I earned my M.S. at Carnegie Mellon as a Fulbright Scholar, where my thesis received the Master’s Thesis Project Graduation Award. Before that, I studied Information Technologies for the Built Environment at the Technical University of Munich as a TEV-DAAD Scholar, and I hold a B.Arch. from Middle East Technical University, graduating as salutatorian. I have also worked on design-technology R&D at KPF and performance analytics tool and workflow development at SmithGroup.
Away from research, I eat (at the very best), drink (specialty coffee, nothing more), sing and play (piano and guitar), and run. In an earlier life, I played competitive volleyball and represented Turkiye in beach volleyball. These days, it is mostly tennis, running, and volleyball when I can.
Buildings are everywhere, but the information needed to understand and improve them is often fragmented across imagery, property records, energy data, and other sources. My work explores how machine learning can bring these incomplete forms of evidence together to understand existing building stock and support retrofit decisions at scale.
