akash7243

Akash Sharma

Hello! My name is Akash Sharma, and I am a Master's (Research) student at IIT Madras currently working as a research associate at the Sudha Gopalakrishnan Brain Centre, IIT Madras where my area of interests include deep learning algorithms for image segmentation, reinforcement learning for generating human-level annotations for medical images, and uncertainty estimation techniques. I was awarded Samsung IIT Research Fellowship for 2022-2023 to pursue research in cyber-physical systems.

I started my foray into AI in the second year of my undergraduate studies when I encountered neural networks for the first time. Since then, I have been incredibly passionate about how such algorithms give rise to intelligence-like behaviors. My curiosity and excitement for this particular field have only increased further as exciting developments occur in the AI research community. During my graduate studies, to share my excitement, I also started a "Journal Club" where like-minded peers discuss the latest topics in CS and AI and hold hands-on deep dive sessions on relevant tools and skills. I have also taken advanced-level mathematics courses like Probability and Applied Linear Algebra along with Deep Learning for Imaging, Reinforcement Learning, and Machine Learning during this time to strengthen my skillset.

I have worked on various projects, such as using deep unrolled networks and recurrent neural networks for power system state estimation & forecasting and from-scratch implementation of reinforcement learning literature like Intra-Option Q Learning & SMDP Q-Learning to solve some famous Gymnasium environments.

Eventually, I want to work at cutting-edge research labs where both research and development skills are required. Thus, I want to gain the necessary skills for collaborative development in time-bound environments. I believe participating in hackathons would be an excellent way to achieve this. I plan to make more exciting projects, such as making chatbots based on chat histories with real people, a general math theorem solver, and AI agents capable of playing games like Dark Souls!

Projects

Virtual GPT

Gamified interaction with some quirky virtual NPCs based on ChatGPT.Unity, Visual Studio, Mixamo, Stability Diffusion, Ready Player Me

Skills

Python
Machine Learning
Deep Learning
Digital Image Processing
Deep Reinforcement Learning

Experience

  • Sudha Gopalakrishna Brain Centre, IIT Madras - Research Associate (Analytics)
    May 2022 - Present