Mo Shahdloo

I am a Senior Applied Scientist in Microsoft's Applied Sciences Group. I develop multimodal agentic systems optimized for efficient on-device inference on Windows.

Previously, I developed machine learning models for markerless motion capture at Vicon and worked on functional MRI acquisition and reconstruction under subject motion at the University of Oxford's FMRIB Centre. My MRI research has explored reconstruction methods combining deep learning, machine learning, and signal processing. My research in computational neuroscience has examined how the brain represents visual and spoken information under natural stimulation conditions by studying fMRI responses using various machine learning and deep learning methods.

I studied electrical engineering at Amirkabir University and Sharif University in Tehran. After developing industrial embedded systems in the energy sector for a couple of years, I moved to Bilkent University in Ankara for graduate research in MRI and computational neuroscience with Tolga Çukur.

More details are in my CV.

Projects

Latent-Lens: visual perception in small language models
A small language model can use visual information from a vision-language model through continuous prefix embeddings. Latent-Lens projects the sender's hidden states into the receiver's embedding space, enabling visual question answering and image captioning. Accepted at NeurIPS 2026.
Project page · Paper

Markerless human pose estimation and tracking
At Vicon, I developed machine learning models that estimate and track human pose directly from video. This approach supports motion capture without attached markers or special clothing.

Motion correction for awake non-human primate fMRI
Movement changes the magnetic field during MRI, making functional imaging of awake monkeys particularly challenging. We developed acquisition and reconstruction methods that correct these changes and improve accelerated fMRI.
2022 paper · 2024 paper

Attention and action representations in the brain
We mapped how hundreds of objects and actions are represented across the cortex, then examined how these representations change when people search for particular action categories.
Paper · Brain viewer

Language, context, and attention in the brain
Using neural language models and brain recordings during story listening, we studied how attention changes the timescales over which the brain integrates context.
Paper

Rapid self-tuning compressed sensing MRI
Compressed-sensing MRI reconstruction depends on regularization parameters that vary across subjects and datasets. We developed a self-tuning method that selects these parameters during reconstruction using projections onto epigraph sets of the l1 and total-variation norms.
Paper

Fat suppression and partial-volume effects in bSSFP MRI
Combining in-phase and out-of-phase acquisitions can suppress fat signals in bSSFP MRI. We studied how the weighting parameter affects fat suppression and partial-volume artifacts, and identified settings for different imaging requirements.
Conference abstract

Publications

Download the bibliography.

Research papers and editorials

Shahdloo M., Sardari F., and Salvaris M. (2026). Latent-Lens: Visual Perception in Small Language Models Through Latent Communication. Advances in Neural Information Processing Systems (NeurIPS 2026, accepted).
Gaspar A., Nagtegaal M., Santini F., Schauman S., Shahdloo M., Van Houdt P. J., Wang Y.-F., and Webb A. (2025). Advancing MRI, together: open science in MR research. Magnetic Resonance Materials in Physics, Biology and Medicine. 38(4). pp 635-638. 10.1007/s10334-025-01286-8.
Priestley L., Chiew M., Shahdloo M., Mahmoodi A., Cheng X., Cleveland R., Rushworth M., and Khalighinejad N. (2025). Dorsal raphe nucleus controls motivation-state transitions in monkeys. Science Advances. 11(26). eads1236. 10.1126/sciadv.ads1236.
Shahdloo M., Khalighinejad N., Priestley L., Rushworth M., and Chiew M. (2024). Dynamic off-resonance correction improves functional image analysis in fMRI of awake behaving non-human primates. Frontiers in Neuroimaging. 3 1336887. 10.3389/fnimg.2024.1336887.
Shahdloo M., Çelik E., Ürgen B. A., Gallant J. L., and Çukur T. (2022). Task-Dependent Warping of Semantic Representations during Search for Visual Action Categories. The Journal of Neuroscience. 42(35). pp 6782-6799. 10.1523/jneurosci.1372-21.2022.
Shahdloo M., Schüffelgen U., Papp D., Miller K. L., and Chiew M. (2022). Model-based dynamic off-resonance correction for improved accelerated fMRI in awake behaving nonhuman primates. Magnetic Resonance in Medicine. 87(6). pp 2922-2932. 10.1002/mrm.29167.
Kiremitçi I., Yilmaz Ö., Çelik E., Shahdloo M., Huth A. G., and Çukur T. (2021). Attentional Modulation of Hierarchical Speech Representations in a Multitalker Environment. Cerebral Cortex. 31(11). pp 4986–5005. 10.1093/cercor/bhab136.
Dar S. U. H., Yurt M., Shahdloo M., Ildiz M. E., Tinaz B., and Çukur T. (2020). Prior-Guided Image Reconstruction for Accelerated Multi-Contrast MRI via Generative Adversarial Networks. IEEE Journal of Selected Topics in Signal Processing. 14(6). pp 1072-1087. 10.1109/JSTSP.2020.3001737.
Shahdloo M., Çelik E., and Çukur T. (2020). Biased Competition in Semantic Representation During Natural Visual Search. NeuroImage. 216 116383. 10.1016/j.neuroimage.2019.116383.
Shahdloo M., Ilicak E., Tofighi M., Saritas E.U., Çetin A.E., and Çukur T. (2019). Projection onto Epigraph Sets for Rapid Self-Tuning Compressed Sensing MRI. IEEE Transactions on Medical Imaging. 38(7). pp 1677-1689. 10.1109/TMI.2018.2885599.
Dar S. U. H., Yurt M., Shahdloo M., Ildiz M. E., and Çukur T. (2018). Synergistic Reconstruction and Synthesis via Generative Adversarial Networks for Accelerated Multi-Contrast MRI. arXiv. 1805.10704.

Conference publications

Shahdloo M., Khalighinejad N., Harbison C., Miller K. L., Rushworth M. and Chiew M. (2022). Dynamic off-resonance correction improves functional data quality in fMRI of awake behaving NHPs. OHBM. Glasgow, Scotland.
Shahdloo M., and Chiew M. (2022). Optimal Singular-Value Shrinkage for fMRI Denoising. ISMRM. London, England.
Shahdloo M., Papp D., Schüffelgen U., Miller K. L., Rushworth M., and Chiew M. (2021). Highly accelerated fMRI of awake behaving non-human primates via model-based dynamic off-resonance correction. ISMRM. Online.
Papp D., Schüffelgen U., Shahdloo M., Rieger S. W., Hess A. T., Rushworth M., and Clare S. (2021). Imaging performance of a multi-channel non-human primate coil. ISMRM. Online.
Shahdloo M., and Çukur T. (2019). Tradeoff between fat-suppression and partial-voluming in weighted combination alternating repetition-time (ATR) balanced SSFP. ESMRMB. Rotterdam, Netherlands.
Shahdloo M., Acar M., and Çukur T. (2019). Attention During Story Listening Modulates Temporal Receptive Windows Across Human Cortex. Cognitive Computational Neuroscience (CCN). 10.32470/CCN.2019.1039-0. Berlin, Germany.
Shahdloo M., Ürgen B. A., and Çukur T. (2019). Attention to action categories shifts semantic tuning toward targets across the brain. OHBM. Rome, Italy.
Dar S.U.H., Yurt M., Shahdloo M., Ildiz E., and Çukur T. (2019). Joint recovery of variably accelerated multi-contrast MRI acquisitions via generative adversarial networks. ISMRM. Montréal, Canada.
Shahdloo M., Ilicak E., Tofighi M., Saritas E.U., Çetin A.E., and Çukur T. (2018). Rapid self-tuning compressed-sensing MRI using projection onto epigraph sets. ISMRM. Paris, France.
Shahdloo M., Ilicak E., Tofighi M., Saritas E.U., Çetin A.E., and Çukur T. (2017). Adaptive Wavelet Thresholding for Profile-Encoding Reconstruction of Balanced Steady-State Free Precession Acquisitions. ESMRMB. Barcelona, Spain.
Shahdloo M., and Çukur T. (2017). Biased competition in semantic representations during category-based visual search. OHBM. Vancouver, Canada.

Open Reviews

I sign my peer reviews and share them here and on PubPeer when a public preprint is available.

Richard H., Martin L., Pinho A.L., Pillow J., Thirion B. (2019). FastSRM: A fast, memory efficient and identifiable implementation of the shared response model . arXiv 1909.12537.
Reply to reviewer comments on: Shahdloo M., Çelik E., Ürgen B. A., Gallant J. L., and Çukur T. (2021). Task-Dependent Warping of Semantic Representations During Search for Visual Action Categories . bioRxiv 10.1101/2021.06.17.448789.

Without music, life would be a mistake ― Friedrich Nietzsche