| 10 min
Tutorial Monitor GPU Training From Anywhere
I needed to monitor my 270M parameter model from the gym. After three failed tunneling attempts, I found a solution that works anywhere, anytime, for free.
Available for opportunities
Pavan Kumar Dubasi
ML Engineer & AI Researcher
NIT Patna. NLP & Continual Learning. Building AI that works in the real world.
About
Passionate about
Passionate about
machine learning.
ML/NLP Focus
Continual Learning
Education
NIT Patna
Experience
2+ Years
Publications
3 Papers
I specialize in building machine learning systems that work in the real world. My research focuses on continual learning and neural machine translation, with a particular interest in preventing catastrophic forgetting in large language models.
Currently training models at home on my RTX 2070 and monitoring them from the gym. Yes, I wrote a whole blog post about that.
Featured Work
Recent Projects
Blockchain
ChainCore
A comprehensive web application showcasing enterprise blockchain infrastructure capabilities. Designed for developers and businesses exploring distributed ledger technology solutions.
React TypeScript Tailwind CSS
Developer Tools
CodeSage
A developer productivity tool leveraging static code analysis and machine learning to enhance software quality. Integrates with development workflows to provide intelligent code review assistance.
Python Flask AST
Game Development
Cosmic Defender
An immersive browser-based space exploration game featuring asteroid navigation, combat mechanics, and power-up collection systems. Demonstrates advanced web graphics programming capabilities.
Three.js JavaScript WebGL
Latest Insights
From the Blog
| 13 min
Career I Cold-Emailed Sony's Research Director. They Called Back. Here's What Happened.
From Principal AI Consultant to internship interview. A technical deep-dive into multimodal AI research, adaptive fusion, and why clarity matters more than titles.
| 12 min
Tutorial I Wasted 2 Hours Daily on GitHub Until I Learned This
I was switching between terminal and browser 50+ times per day. Then I discovered GitHub CLI. Time saved per day went from 2 hours to 5 minutes.
Contact
Let's build
Let's build
something together.
Open to research collaborations, interesting projects, and conversations about machine learning.