Alok
I enjoy turning research into friendly tools, from agent workflows to applied ML systems, and I love reading more about tranformers, LLM training and inference optimization .
Highlights
Public repos—
Followers—
projects & activity
pinned repositories for aalok-p.
getting there...
Experience
- Machine Learning Intern @ AlaricTech [oct 25' - dec 25']
- AR Developer @ GDG KIET [jan 24' - jan 25']
- Intern @ TBI KIET [may 24' - june 24']
- Gaudmire @ buildspace [may 24' - july 24']
Projects
- MEERA - AI Companion
- it isn't just another chatbot. It's an emotional companion made to truly listen and remembers about user.
- uses fine tunned Qwen3-32B with qlora adapters for empathetic dialogue.
- BrightWalk
- visual assistant for impaired person, it describes what it sees using person's pov.
- mobile application built for visually impaired to help me naviagte and do their task easily.
- SAR image Colorization
- uses conditional GAN combining a ResUNet generator and PatchGAN discrimator to convert noisy sar(grey) images into colored rgb images.
- have multi-term objective (L1 reconstruction loss with weight 100) to stabilize supervised SAR–RGB translation.
- Pratham
- a custom CNN to classify MRI scans for brain tumor detection with high accuracy.
- integrated grad-cam visualizer for easier interpretion, with 2d-1d transformation pipeline.
- AI-Code
an open-source project designed to help individuals learn and understand fundamental code implementations of various AI algorithms
things I am learning
you will find me learning LLM training, inference optimization and a lot of kernel code, also learning about what a frontier labs require, If I am not learning and creating then you can find me in 9-5 lec (ps- sed emoji).