> whoami

& Full-stack Engineer

Incoming 4th-year CS student who loves AI. I am currently creating my own ML framework (for learning purposes) and working on an app that can help beginners start static calisthenics--injury free

> about_me.txt

I'm currently pursuing my Bachelor's degree in Computer Science with a focus on artificial intelligence, distributed systems, and databases.

When I'm not programming, you can find me reading research papers, learning something new, hiking mountains, watching anime, practicing calisthenics, or cycling (on a good day)

Focus Areas

Machine Learning & Deep Learning

LLMs, RAG Systems & Finetuning

System Design, APIs, & Relational Databases

> tech_stack.json

Python

Java

JavaScript/TypeScript

React

PyTorch

LangChain

PostgreSQL

MySQL

Django

FastAPI

Docker

Kubernetes (GKE)

AWS

GCP

> experience.log

Undergraduate Machine Learning Research Intern

Michael Smith Labs, University of British Columbia

Jan 2025 - Present•Vancouver, BC

Collaborating with an interdisciplinary team to research neural behavior decoding using machine learning
Fine-tuning a self-supervised neural decoding model using Python and PyTorch to boost interpretability and performance

Python

PyTorch

NumPy

Matplotlib

BMTK

Simulations

Synthetic Data

Software Engineering Intern

Ducem Technologies

May 2024 - Aug 2024•Surat, GJ

Improved data load times by 35% for a critical product used by 50,000+ users, through progressive loading, caching, and API response compression
Took initiative to resolve an API bug in an ASP.NET product, reducing client-side processing delays from 13s to 2s
Implemented a unit-test suite using JavaScript and Jest, preventing 7 regressions before the next production push

JavaScript

TypeScript

Angular

ASP.NET

MySQL

Supabase

Azure

Jest

> education.json

Bachelor of Science in Computer Science

University of British Columbia, Vancouver

Sep 2022 - May 2027 (Expected)

Relevant Coursework

Data Structures & Algorithms

Applied Machine Learning

Introduction to Relational Databases

Introduction to Software Engineering

Introduction to Artificial Intelligence

Activities & Leadership

UBC Open Robotics

Brock Commons Residence Association Tallwood President (2025-26)

UBC BizTech

UBC CSSS

> selected_projects/

Thirdslate: AI-Powered Tutoring Platform

Problem I aimed to solve: Wanted to fix how ChatGPT gives generic answers, so I built Thirdslate, an AI tutor that teaches you like a friend who knows your level, remembers your struggles, and grows with you

Solution: Architected 3 end-to-end agentic RAG tutoring systems with custom LangGraph orchestrations for personalized tutoring. Coded a scalable backend. Established CI/CD pipeline using GitHub Actions. Attained a 4.8/5 rating for tutoring quality and satisfaction. Achieved 99.95% uptime by containerizing backend using Docker and deploying on Google Cloud Platform (GCP) with horizontally autoscaling Kubernetes clusters

Python

React

NextJS

PostgreSQL

Supabase

REST API

FastAPI

Uvicorn

Docker

Kubernetes

GCP

CI/CD

RAGs

LangGraph

Demo

Ecocircle: Circular Economy Marketplace App (AWS Hackathon: 3rd place)

Problem I aimed to solve: Noticed the rise of sustainability trends and built a platform for eco-conscious users to easily buy and sell eco-friendly products

Solution: Developed a web marketplace for sustainable commerce and upcycling. As the lead on the AI and AWS architecture, I was responsible for integrating GenAI & ML, engineering the backend architecture and deploying it across a scalable AWS infrastructure

Python

API Gateway

AWS Lambda

AWS Bedrock

DynamoDB

Code

Transformer From Scratch

Problem I aimed to solve: Wanted to learn how modern systems like ChatGPT work under the hood

Solution: Implemented the Transformer architecture from scratch introduced in the Attention Is All You Need paper, built using PyTorch without relying on high-level libraries

Python

PyTorch

NumPy

Pandas

Matplotlib

Code

Conception

Problem I aimed to solve: To help new architects, students, or curious learners like me overcome the complexity of traditional architecture software

Solution: Created a desktop app that enables homeowners to seamlessly navigate through the home design process without the need for extensive technical knowledge

Java

Java Swing

JUnit Testing

Advanced OOPs

Code

Calisthenics Form Feedback System

Problem I aimed to solve: Since I was just starting calisthenics, I didn't have anyone to correct my form and bad form in calisthenics can lead to serious injuries

Solution: Wrote a real-time calisthenics form feedback system using MoveNet, achieving performance tracking at 200 FPS

Python

Tensorflow

MoveNet

Code

Cropwise

Problem I aimed to solve: Most farmers often can't tell if a plant is diseased or how serious it is until it's too late

Solution: Built a hybrid CNN-based leaf disease detection model, trained for 50 epochs, achieving an accuracy of 90%

Python

Tensorflow

Keras

Albumentations

SMP

Code

> contact.sh

I'm always open to discussing new opportunities, collaborating on interesting projects, or just having a chat about technology and innovation