Atish Dixit

Atish Dixit

Research Fellow

University of Edinburgh

Atish Dixit is a Research Fellow at the APRIL AI hub, University of Edinburgh, developing AI-driven approaches for discovering and characterising electronic materials (with collaboratory PIs: Prof. Themis Prodromakis, Prof. Jacqui Cole, and Prof. Subramanian Ramamoorthy). His work brings together machine learning, mathematical modelling, simulation, and experimental insight to build more interoperable and trustworthy scientific-discovery workflows.

His research spans deep and multi-agent reinforcement learning, robust optimal control, computational physics, and materials informatics. He is particularly interested in making AI research useful beyond the model itself: reproducible methods, transparent evaluation, and collaborations that connect data, physical systems, and real-world decisions. Before academia, Atish worked in engineering roles across energy and automotive applications, grounding his research in practical modelling problems.

Interests
  • Artificial Intelligence
  • Materials Discovery
  • Multi-agent Robotic Systems
  • Deep Reinforcement Learning
  • Computational Physics
Education

  • PhD in Deep Reinforcement Learning, 2023

    Heriot Watt University

  • MTech in Mathematical Modelling, 2017

    University of Pune

  • BEng in Mechanical Engineering, 2012

    University of Pune

Skills

Technical
Programming
Machine Learning and Statistics
Mathematical Modelling and Computing
Hobbies
Study of Vedanta
Painting
Boxing

Experience

 
 
 
 
 
University of Edinburgh
Research Fellow
University of Edinburgh
December 2024 – Present Edinburgh, UK
AI-driven materials discovery for electronic materials
 
 
 
 
 
University of Edinburgh
Research Associate
University of Edinburgh
October 2023 – October 2024 Edinburgh, UK
State-of-the-art research in multi-agent reinforcement learning
 
 
 
 
 
Heriot-Watt University
PhD Researcher
Heriot-Watt University
November 2018 – March 2023 Edinburgh, UK
Research in deep reinforcemnt learning methods to solve opimal control problems in computational physics
 
 
 
 
 
ExxonMobil
Computational Engineer
ExxonMobil
March 2017 – September 2018 Texas, US / Bangalore, India
  • Mathematical modelling and back-end development of on-field applications
  • Analysis and process recommendation for on-field drilling process
  • Part of agile software development for various reservoir management tools
 
 
 
 
 
Indian Institute of Technologies, Madras
Research Fellow
Indian Institute of Technologies, Madras
May 2016 – July 2016 Madras, India
  • Engaged in research and development within the domain of Graph Theory
  • Formulated a C++ object-oriented design to model graphs and implement the Christofides algorithm
 
 
 
 
 
Tata Technologies
Engineering Analyst
Tata Technologies
July 2012 – August 2015 Pune, India
  • Design recommendation and analysis of automobile system architectures for HVAC
  • Building reports of design failures/improvements using computational modelling of vehicle aerodynamics
  • Research and development of novel mathematical models to assess human comfort in automobile cabins

Recent Posts

When Materials Discovery Speaks in Different Languages
Digital materials discovery has never had more powerful tools, yet disconnected databases, models, and workflows still slow the search for useful electronic materials.
Atish Dixit
Sep 16, 2025 4 min read
When Materials Discovery Speaks in Different Languages
Understanding Advaita Vedanta
My perspective and understanding of Advaita Vedanta
Atish Dixit
Aug 13, 2024 3 min read
Understanding Advaita Vedanta
Reproducibility in Deep Reinforcement Learning Literature
A post on importance of reproducibility in reinforcement learning literature using example of my paper on multilevel PPO
Feb 4, 2024 3 min read
Reproducibility in Deep Reinforcement Learning Literature
Tales from the PhD Trenches
Navigating a PhD is like riding a wild rollercoaster, so embrace humor, resilience, and camaraderie to thrive amidst the chaos. Celebrate small victories, avoid procrastination, and remember that failure is just another step in your grand experiment.
Atish Dixit
Dec 13, 2022 2 min read
Tales from the PhD Trenches
When Academia Meets Industry
The Edinburgh SIAM-IMA student chapter orchestrated a collaborative week-long event where academia and industry synergized to solve real-world challenges, demonstrating the transformative potential of interdisciplinary cooperation in mathematical applications.
Atish Dixit
Aug 13, 2021 2 min read
When Academia Meets Industry

Projects

*
All Deep Learning Other
Electronic Materials Ontology System
An open-source framework for connecting databases, machine learning models, and workflows for electronic materials discovery
Electronic Materials Ontology System
Reinforcement Learning Sample Efficiency Improvement using Multi-Grid Approach in Computational Physics Problems
A multi-grid approach in deep reinforcement learning to improve sample efficiency in computational physics based applications
Reinforcement Learning Sample Efficiency Improvement using Multi-Grid Approach in Computational Physics Problems
Deep Reinforcement Learning for Reservoir Management
An RL-based approach to learn the optimal well control policies used to improve the production efficiency in reservoir management
Deep Reinforcement Learning for Reservoir Management
Multilevel Deep Reinforcement Learning Framework
A statistical framework to reduce computational complexity of RL applications in computational physics based environments
Follow
Multilevel Deep Reinforcement Learning Framework
Mathematical Modelling of Human Bioheat Transfer
A mathematical model of human bio-heat equation and assessment of human cumfort in various environment conditions
Mathematical Modelling of Human Bioheat Transfer

Artwork

468B017D-BCD7-4A12-BFA5-8B3AF3FA6164.jpg
CABFDB3C-DF5B-4968-B528-17707204864A.jpg
IMG_0672.JPG
IMG_0720.JPG
IMG_2630.JPG.jpg
IMG_4637.JPG
IMG_6555.jpg
IMG_8685.JPG
IMG_8863.JPG

Recent Publications

Quickly discover relevant content by filtering publications.
Cristian Sestito, Panagiota Kontou, Pratibha Verma, Atish Dixit, Alexandros D. Keros, Michael O'Boyle, Christos-Savvas Bouganis, Themis Prodromakis (2026). A flexible language model-assisted electronic design automation framework. arXiv preprint arXiv:2601.14098.

PDF Cite

Atish Dixit, Alexandros D. Keros, S. Mishra, B. Rowlinson, J. Cole, S. Ramamoorthy, Themis Prodromakis (2026). Unifying Digital Discovery of Electronic Materials with EMOS. Research Square preprint.

PDF Cite DOI

Atish Dixit, Ahmed H Elsheikh (2023). Robust optimal well control using an adaptive multigrid reinforcement learning framework. Mathematical Geosciences.

Cite

Atish Dixit, Ahmed Elsheikh (2022). A Multilevel Reinforcement Learning Framework for PDE-based Control. arXiv preprint arXiv:2210.08400.

Cite

A Dixit, A Elsheikh (2022). Robust Well-Production Control Using Surrogate Assisted Reinforcement Learning. ECMOR 2022.

Cite

See all publications

Contact

For any inquiries or collaborations, feel free to reach out to me via the contact form below.

  • adixit@ed.ac.uk
  • 10 Crichton St, Newington, Edinburgh, Scotland EH8 9AB
  • Enter Informatics Forum, register to your entry and take the stairs to Office 2.43 on Floor 2
  • Monday to Friday 09:00 to 17:00
  • Book an appointment