GHESHGHEREGH@GMAIL.COM
|
About
I am a Researcher at Cohere. My research is focused on advancing the frontiers of generative AI and Machine Learning, particularly reinforcement learning and self-supervised learning. My ultimate goal is to build the super-intelligent AI capable of solving the problems that humans struggle to solve, such as understanding the origin of universe, finding the unified theory of physics, mastering space travel, curing cancer, scaling up clean energy (also eliminating tin-pot dictatorships!). |
2016 - 2023 | London, UK
2023 | Seattle, WA
As a Staff Research Scientist at DeepMind, I spearheaded a team of talented research scientists and engineers in developing multiple ground-breaking AI technologies. This journey resulted in key advancements in areas like fine-tuning generative AIs with AI-feedback, self-supervised learning, world models, exploration, and diffusion models. Notable projects include BLaDE, BYOL-Explore, and World Discovery Models, among others.
My contributions extended from the project ideation phase through to execution, leading to high-impact publications and patents. Some significant projects in this regard include Noisy Networks for Exploration, Neural Belief Representations, and Minimax RL.
Throughout my tenure, I had the privilege of collaborating with leading researchers on influential projects such as computer control with generative AI and Bootstrap-Graph representation Learning. This collaborative effort led to multiple well-cited publications, contributing to the broader AI and machine learning community.
As a Postdoctoral Fellow, I contributed to the publication of several peer-reviewed scientific papers for prestigious conferences such as NeurIPS and ICML. Notable works include 'A cryptography-based approach for movement decoding' and 'Convex Relaxation Regression'.
Radboud University Nijmegen | Nijmegen, Netherlands | October 2012
University of Tehran | Tehran, Iran
K N Toosi University of Technology | Tehran, Iran