Projects
A selection of my recent work in fullstack development, artificial intelligence, and computational science.
SecondMind
AI-powered learning platform that transforms topics or documents into interactive roadmaps and actionable tasks powered by Google Gemini API to dynamically generate highly structured conceptual trees and contextual to-do lists from user prompts. Built a dynamic frontend canvas utilizing React Flow and Dagre for algorithmic graph layouting, and configured Vercel serverless cron jobs to dispatch automated daily email reminders via Resend.
Personal Website
Dynamic personal portfolio website with a custom-built content management system, allowing live updates of portfolio entries via restricted server-side actions and Cloudinary integrations. Implemented robust backend operations for Cloudinary image uploads and a spam-resistant Nodemailer contact system utilizing Cloudflare Turnstile CAPTCHA and hidden honeypots.
Bite-sized Coding
Gamified typing practice for developers on dynamically-fetched snippets from public repos of their chosen language using GitHub API, as well as an on-demand code explanation feature powered by Gemini. Uses a custom React typing interface to simulate typeracer-like experience.
This or That
Real-time web implementation of “This or That” with private rooms and custom question prompts and responses powered by Gemini. Engineered a real-time multiplayer Node.js backend using Socket.io, implementing an in-memory state machine to synchronize complex game phases across clients.
Masking-Augmentation on Mel-Spectrogram and MFCC for White-crowned Sparrow Vocalization Variants Classification
A CNN-based audio classification system using Mel-spectrogram and MFCC features for vocalization variants of White-crowned Sparrows. Utilized frequency and time axis masking for data augmentation to achieve 98% F1-Score, outperforming non-augmented models. Presented at ISRITI 2025.