Shubh Jani

About Me

Combining technical expertise with social entrepreneurship to create meaningful change

I'm a high school senior working on leveraging computer science and engineering to solve pressing challenges in:

My work spans machine learning research, full-stack development, and social entrepreneurship with presentations at the UN and research at national labs.

Skills & Expertise

Technical skills and research focus areas

Technical Skills

Transformers & Diffusion Models
- Developed QwenVL-Braille vision-language model at OpenNLP Labs, creating an AI agent for Braille education using transformer architectures.
LSTM in RNNs, Reinforcement Learning
- Used LSTM networks for multivariate time series analysis at Berkeley EINR to optimize data cache utilization patterns. Utilized reinforcement learning to optimize autonomous cars for energy efficiency.
Node.js, React, PostgreSQL
- Built full-stack platforms with backend services such as ItsClean and AarogyaVardaan to facilitate users accessing health benefits or reviewing bathroom cleanliness.
AWS, OnShape, PCB Design
- Designed and deployed cloud infrastructure with AWS in ItsClean; hardware prototypes for Celeritas Research using OnShape; PCB design in autonomous cars for RoboRacer competition.

Research Focus

Low-resource ML applications
- Created personalized AI agent for Braille education targeting underprivileged communities with limited resources.
Neural network optimization
- Optimized neural networks through hyperthreading and sparse compression techniques (4th place at Alameda County Science Fair).
Vision-language models
- Developed QwenVL-Braille model combining visual and language processing for accessibility applications.
Ethical AI development
- Created Fairify app to recognize bias in ML datasets, winning NYAS Ethical AI competition.