Vihaan Sharma

My coding journey began in grade 5, when I discovered I could edit website text with Inspect Element — and prank my friends. By grade 9, I was deep into freeCodeCamp, learning HTML and CSS, and I’ve been creating projects ever since. My Coding folder might be a chaotic mess, but it’s home to everything from fun experiments to polished web apps — some of my favorites are featured above. Outside of coding, I’m into TV shows, anime, AI, badminton, and working out. This fall, I’m starting my undergrad at the University of Toronto, and I’m excited for what’s next.

Projects

ChessBrain

Client

https://chessbrain.org

ChessBrain is a comprehensive chess training platform I developed using Next.js, TypeScript, and Supabase. The application allows chess enthusiasts to improve their skills by playing through positions from grandmaster games and comparing their moves to those of top players. I implemented a fully interactive chessboard with move validation, detailed annotations, and variation analysis. The platform features a freemium subscription model using Stripe, comprehensive user progress tracking, and theme customization options. The project demonstrates my ability to create complex, interactive web applications with sophisticated state management and third-party integrations.

NextJS
Tailwind CSS
React
Supabase
NextUI

wardrobe.ai

Hackathon

https://github.com/PconnorsUWO/HT6-Team

Wardrobe.ai is a browser‑based, AI‑powered virtual try‑on that lets shoppers see garments on their own body in near‑real time—no app install, just open the camera and go. Users chat intent via an interview flow (Ribbon), step into frame, and the system auto‑captures an optimal pose; IDM‑VTON then warps and blends the selected garment onto the user’s photo with photorealistic drape and alignment. The frontend (React + TypeScript on Vite with Tailwind and shadcn/ui) streams frames and receives results over Socket.IO for low‑latency feedback, while a Python/Flask backend orchestrates pose/landmark detection, try‑on processing, and result delivery. Smart suggestions powered by Vellum complement the look. The result is a mirror‑like try‑on that boosts confidence, reduces returns, and works consistently across mobile and desktop. Core flow: chat intent → center in frame → auto‑snap → instant try‑on → complementary recommendations AI/CV: MediaPipe landmarks, OpenCV pre/post‑processing, IDM‑VTON for cloth warping Frontend: React, TypeScript, Vite, Tailwind, shadcn/ui, Socket.IO Client Backend: Flask + Flask‑CORS, real‑time via Socket.IO; storage/DB not required for core demo flow

NextJS
Tailwind CSS
Vellum AI
TypeScript
Socket.IO Client
Flask
OpenCV

DermAI

Application

https://github.com/RishiShah99/DermAI

DermAI is an AI-driven dermatology assistant that leverages deep learning to classify skin conditions, assess urgency, and provide actionable guidance. Built with a modern Next.js + TypeScript frontend and a Python Flask backend, DermAI uses a fine-tuned DenseNet121 PyTorch model trained on the DermNet dataset (19,500 images, 23 categories) to deliver accurate, near-instant predictions. The frontend integrates the Vercel AI SDK with Google's Gemini models for an interactive chatbot experience, while Supabase handles image storage and document metadata. A separate backend service manages ML inference, supported by RAG utilities for embedding-based retrieval from dermatology knowledge sources. Key features: Fast AI classification: Upload or capture an image for real-time skin condition prediction (4–5 seconds). Urgency detection: Categorizes cases as urgent, moderate, or non-critical. Interactive guidance: AI chatbot suggests next steps and when to seek medical help. RAG-powered knowledge lookup: Uses OpenAI embeddings for relevant document retrieval. Supabase integration: Stores patient images and related metadata securely. DermAI reduces misdiagnosis rates, improves access to dermatological care in underserved areas, and lowers healthcare costs by filtering benign cases before specialist visits. With a Top-1 accuracy of 66% and Top-5 accuracy of 93%, it matches or outperforms human dermatologists in many image-based diagnostic tasks. DermAI represents the future of accessible dermatological healthcare, with planned expansions into wearable integration, telemedicine partnerships, and multi-language support.

NextJS
TypeScript
Tailwind CSS
Framer Motion
Supabase
Flask
Pytorch
OpenAI

Projects
About
Contact

Hi, I'm

📍 Toronto, UofT CS

Full-stack developer in Canada. I turn ideas into smooth, high-performance web experiences that people actually enjoy using.

Hover a project from the list

Projects

+ way more on my github

Tap a project to open the gallery.

For Clients...

I love working closely with clients, making sure we're always on the same page with clear, honest chats.

I'm very flexible with time zone communications

I constantly try to improve

My Tech Stack

Next.js

React.js

Typescript

Node.js

Express.js

PostgreSQL

Next.js

React.js

Typescript

Node.js

Express.js

PostgreSQL

Next.js

React.js

Typescript

Node.js

Express.js

PostgreSQL

Next.js

React.js

Typescript

Node.js

Express.js

PostgreSQL

I've been coding since Grade 8 and still get a kick out of diving into new projects.

My Technologies

+ much more...

Ready to build something awesome together?
Let's get started!

Ready to take your project
to the next level?

Contact me today, and let's talk about how I can assist you in reaching your goals, or if you just want to chat :)
In any of the following ways:

Hi, I'm

Projects

For Clients...

I love working closely with clients, making sure we're always on the same page with clear, honest chats.

I'm very flexible with time zone communications

My Tech Stack

I've been coding since Grade 8 and still get a kick out of diving into new projects.

My Technologies

Ready to build something awesome together? Let's get started!

Ready to take your project to the next level?

Ready to build something awesome together?
Let's get started!

Ready to take your project
to the next level?