Majid Rahmanov

I believe I deserve the KIVI EngD Award because I translated a demanding fusion diagnostic  

problem into a working software prototype, connecting machine learning with the constraints of  

experimental hardware. At DIFFER, in collaboration with Eindhoven University of Technology, I  

took responsibility for the engineering path from understanding stakeholder needs to  

implementing and evaluating GPU reconstruction.

The challenge was to reconstruct where light originates in the plasma exhaust of MAST-U quickly  

enough to support future control. The project's existing iterative methods took approximately 50 to  

80 milliseconds per frame, against a three-millisecond reconstruction budget. I combined  

calibrated camera geometry with a U-Net and trained it on synthetic emission patterns, including  

a check that reconstructed maps reproduce the camera measurements. My contribution connects  

this modelling approach to data generation, model export, benchmarking and native C++  

integration.

The results demonstrate both technical progress and engineering judgement. The best evaluated  

model achieved structural similarity of 0.930 on held-out synthetic data. A smaller development  

configuration averaged 2.31 milliseconds per frame. The full-camera A100 C++ prototype  

averaged 5.82 milliseconds across six frames, establishing native execution while identifying the  

remaining optimisation gap. I documented these distinct results and their limits so that  

subsequent development can target the actual bottlenecks. Broader experimental validation and  

sustained performance testing remain necessary before use in live control.

I also took ownership of decisions that connected disciplines. When the initial sparse  

backprojection implementation could not be exported through the selected toolchain, I adapted  

the architecture to a GPU-resident dense operation and evaluated the resulting memory and  

performance trade-offs. I worked with DIFFER scientists and engineers and my TU/e supervisor  

to align reconstruction quality, runtime interfaces and hardware constraints. I incorporated  

scientific feedback from UKAEA and EPFL collaborators. This required explaining design choices  

across specialties and maintaining a shared understanding of what the evidence established.

The immediate value is a modular, documented reconstruction pipeline that DIFFER can extend  

without rebuilding the entire development process. Its longer-term potential is to support faster  

plasma exhaust diagnostics and research into protecting components in future fusion devices.  

The work contributes a concrete engineering step towards that goal, with a clear route for further  

validation.

This project reflects the contribution I want to make as an engineer: turning advanced  

computational methods into useful scientific tools, taking responsibility for implementation and  

giving collaborators evidence they can build on. It brings together practical innovation, a reusable  

result for fusion research and collaboration across physics, software engineering and highperformance computing