Gyanateet Dutta |
Master's student at University of Leeds specializing in cutting-edge AI, computer graphics, and quantum technologies. Passionate about pushing the boundaries of what's possible in digital innovation.
About Me
Passionate about the intersection of AI, graphics, and quantum computing
Hi there! I'm Gyanateet, a passionate researcher and developer at the forefront of artificial intelligence and computer graphics. Currently pursuing my Master's degree in Computer Science & AI at the University of Leeds, I'm dedicated to exploring the fascinating intersection of machine learning, quantum computing, and visual technologies.
My journey in technology has led me through various exciting domains - from developing cutting-edge neural networks for image generation to researching quantum algorithms for optimization problems. I'm particularly fascinated by how we can leverage AI to create more immersive and intelligent digital experiences.
When I'm not diving deep into research papers or coding the next breakthrough algorithm, you'll find me exploring the latest advancements in GPU computing, experimenting with generative AI models, or contributing to open-source projects that push the boundaries of what's possible in computer graphics and machine learning.
Education
Master's in CS & AI
University of Leeds
Achievement
8th Place in JAX Diffusers
Hugging Face Competition
Publications
Research papers in CV
& Optimization
Technical Skills
My expertise across various technologies and domains
Featured Projects
A showcase of my most impactful work in AI and computer graphics
LoRA Dreambooth
LoRA adaptation weights for Stable Diffusion XL, trained on anime finetuning dataset with varied art styles for high-quality image generation.
JAX Diffusers ControlNet
8th place in Hugging Face JAX Diffusers competition. Created a ControlNet for anime-realism art generation using JAX and TPU v4 MXUs.
Enhanced Pothole Detection
Novel algorithm combining YOLOv7 with ESRGAN for improved pothole detection using low-resolution cameras. Published research with practical applications.
Quantum TSP Solver
Solving the Travelling Salesman Problem using Hopfield Neural Networks and Simulated Annealing. Comparative study of classical and quantum approaches.
Interactive Portfolio
Modern, interactive portfolio website with 3D graphics, smooth animations, and responsive design. Built with cutting-edge web technologies.
RL Game Agent
Reinforcement learning agent trained using PPO algorithm to play classic games. Demonstrates advanced RL techniques and game AI development.
Gemma-Le VLA (Robotics)
Compact vision-language-action policy for robotic manipulation using LeRobot with SigLIP vision and Gemma 3 for language reasoning.
A2C on PandaReachDense-v2
Actor-Critic agent trained on PandaReachDense-v2 with vectorized environments and normalized observations.
PPO (CartPole/LunarLander)
Proximal Policy Optimization agent with training logs and evaluation on classic control tasks.
Q-Learning (FrozenLake 4x4)
Tabular Q-learning agent trained on no-slippery FrozenLake with hub packaging and usage instructions.
Dalton Mills: VR & AI Preservation
University of Leeds HELIX case study: reconstructing Dalton Mills with deep learning-driven 3D techniques and VR.
Aesthetics of AI Reconstruction
Open collaboration exploring ethical and aesthetic implications of AI-driven reconstruction workflows.
Media Showcase
Demo videos from my models and competitions
Career Timeline
Key milestones in my academic and professional journey
JAX Diffusers Competition
2024Achieved 8th place globally in the Hugging Face JAX Diffusers competition, developing anime-realism ControlNet using TPU v4 architecture.
Published Research Paper
Jan 2024Published "Improved Pothole Detection with YOLOv7 and ESRGAN" on arXiv, introducing novel super-resolution techniques for computer vision applications.
Master's Degree
2023 - PresentCurrently pursuing Master's in Computer Science & AI at University of Leeds, specializing in computer graphics, machine learning, and quantum computing.
Quantum TSP Research
Feb 2022Published research on solving TSP using Hopfield Neural Networks and Simulated Annealing, contributing to quantum optimization algorithms.
Open Source Contributions
2021 - PresentActive contributor to various open-source projects in AI/ML, with focus on computer graphics and machine learning frameworks.
Research & Publications
Contributing to the advancement of AI and computer science
Improved Pothole Detection with YOLOv7 and ESRGAN
This paper proposes a unique algorithm to use low-resolution cameras for automatic pothole detection using super-resolution through SRGANs, establishing a baseline performance with the YOLOv7 network for practical road infrastructure monitoring.
Solving TSP using Hopfield Neural Networks and Simulated Annealing
A comprehensive study on the Travelling Salesman Problem, addressing it with Hopfield Neural Networks and comparing performance to other optimization algorithms for solving graph theory problems in computational complexity research.
Anime-Realism Art Generation using JAX and TPU Optimization
Development of an advanced ControlNet model for anime-realism art style generation using JAX framework optimized for TPU v4 MXUs. Achieved 8th place globally in the Hugging Face JAX Diffusers community competition.
Get In Touch
Ready to collaborate on exciting projects or discuss opportunities