← Back to portfolio

Garvit Joshi

Doctoral Researcher & Research Associate

Machine Learning · Neuroimaging · Interpretable AI · Scientific Software

Dresden, Germany · garvit.joshi@tu-dresden.de · garvitjoshi.me
GitHub · LinkedIn

Doctoral researcher in Systems Neuroscience with experience spanning interpretable machine learning, statistical modelling, neuroimaging, medical-image analysis, geospatial data, and autonomous systems. I develop reproducible analysis workflows and scientific software that make complex models and results easier to understand and use.

Core expertise

  • Machine learning & data: Python, PyTorch, TensorFlow, deep learning, graph neural networks, statistical modelling, BERT/BioBERT
  • Neuroimaging: fMRI, structural MRI, connectivity analysis, FreeSurfer, SPM, medical-image segmentation
  • Software & infrastructure: Git, CI/CD, Linux, CUDA/HPC, Django, Flask, Jupyter, SQL, scientific visualisation
  • Logic & knowledge: Computational logic, ASP, Prolog, OWL, SPARQL, ontology engineering

Professional experience

Research Associate & Doctoral Researcher — Technische Universität Dresden

November 2022–present · Doctoral research in Systems Neuroscience since September 2023, embedded in TRR 265.

  • Study relationships between functional and structural MRI, human behaviour, and personality using machine learning and statistical modelling.
  • Develop interpretable, reproducible workflows for complex neuroimaging models and maintain research-oriented software.
  • Preregistered analysis plans on OSF to strengthen transparency and reproducibility.
  • Contributed to ConnExplorer, linking brain-connectivity matrices with anatomical regions and network-level visualisations.
  • Doctoral work funded through ESF Plus and the Saxon Landesinnovationspromotion through August 2027, alongside a part-time research contract.

Research Assistant, Vehicle Control & Sensor Systems — Fraunhofer IVI

June–October 2022

  • Contributed to helyOS, an open-source platform for coordinating autonomous vehicles.
  • Adapted software for autonomous lawnmowers, route planning, and multi-vehicle operation.
  • Supported systems integration and produced technical documentation.

Research Assistant, Geospatial Data Analysis — Fraunhofer IVI

April 2021–March 2022

  • Analysed road-accident data and contributed to a classifier for identifying hazardous road segments.
  • Applied geospatial statistics and developed an interactive Django mapping application.
Garvit Joshi · Curriculum Vitae 1 / 2

Education

MSc Computational Logic — Technische Universität Dresden

September 2018–April 2022

  • Interdisciplinary programme in artificial intelligence, formal logic, mathematics, and computer science.
  • Master’s thesis at NCT Dresden comparing weakly supervised deep-learning approaches for medical-image segmentation.
  • Project comparing BERT and BioBERT with MetaMap for diagnosis classification from patient records.

BE Computer Science and Engineering — CDGI / RGPV, India

July 2014–July 2018

  • Engineering foundation in software development, AI, databases, networks, algorithms, and embedded systems.
  • Bachelor’s thesis on MIDI music generation using recurrent neural networks and LSTMs.

Selected research & projects

  • Interpretable Neural Networks for the Analysis of Brain Functions — TRR 265: Doctoral research on interpretable modelling of neuroimaging, behaviour, and personality.
  • ConnExplorer: Interactive research software connecting brain-connectivity matrices, anatomical regions, and network representations.
  • SFB 940 & Project Z2 Core Imaging Facility: Earlier work supporting volition and cognitive-control research through deep-learning analysis, scientific software, and neuroimaging infrastructure.
  • VR brain-research exhibit: Virtual-reality science-communication project exhibited at the German Hygiene Museum Dresden.

Open science

Improving Interpretability of Complex fMRI Models of Human Behavior — OSF preregistration documenting hypotheses and analysis plans for part of the doctoral research programme.

Earlier innovation

  • Microsoft Imagine Cup international finalist: Co-developed an application to support people with Alzheimer’s disease and assess memory.
  • Helios: Contributed to a Raspberry Pi assistive device combining ultrasonic sensing, computer vision, and hardware integration; presented at Startup Istanbul in 2017.

Languages

Hindi — native · English — professional working language · German — Goethe B1 certified and C1.1 academic course completed · Japanese and Czech — basic knowledge