ProjectsExperienceResearchContact
open to SDE-1 roles

I turn ideas into
working software.

Software Developer

Full-Stack Developer building useful web products, AI-enabled tools, and connected experiences. My IEEE-published Vision Transformer research reached 94.5% accuracy on FaceForensics++.

View Projects
4+
Projects shipped
1
IEEE paper published
350+
DSA problems solved
02 — Work

Selected Projects

Software built across full-stack workflows, AI-assisted tools, and carefully crafted interfaces.

7 projects

Featured · AI/Tools

AI Data Cleaner & Visualizer

A full-stack data-cleaning tool built with Next.js, TypeScript, and Gemini API. It turns text, CSV, XLSX, and PDF inputs into validated, typed datasets, with type inference, deduplication, and ready-to-run SQL exports.

ReactNext.jsTypeScriptGemini APICharts
Live Demo Code
Full-Stack

Eraser.io Clone

A responsive recreation of the Eraser.io diagramming interface, with an interactive canvas and custom drawing tools. Focused on pixel-level detail and smooth, familiar interactions.

ReactCanvas APITailwind CSS
Live Demo Code
Full-Stack

Taskly — Project Task Management

A task-management platform built with Next.js, Tailwind CSS, Node.js, Express, and MongoDB. Full CRUD workflows support creating, assigning, updating, and tracking project tasks.

Next.jsNode.jsExpressMongoDB
Live Demo Code
AI/Tools

AI Assistant — Chrome Extension

An LLM-powered Chrome extension that brings translation, code generation, and contextual help into the current web page, using Chrome Extension APIs.

JavaScriptChrome-APIsLLM API
Code
Frontend

E-Commerce Website

A responsive product-browsing experience built with React and Tailwind CSS, using the DummyJSON API. The work focused on reusable components and a consistent layout across screen sizes.

React.jsTailwind CSSDummyJSON API
Live Demo Code
Frontend

QR Code Web App

A lightweight QR-code generator built with HTML, CSS, JavaScript, and the QR Server API, with a simple responsive interface for turning text into a code.

HTMLCSSJavaScriptQR Server API
Live Demo Code
AI/Tools

DeepFake Image Detection

A deployable DeepFake detection pipeline using ViT-B/16, Python, PyTorch, Hugging Face Transformers, Flask, and FastAPI. The model reached 94.5% accuracy on FaceForensics++.

ViT-B/16CNNsPythonPyTorchTensorFlowHugging FaceFlaskFastAPI
Code
03 — Journey

Experience & Education

Product work, freelance builds, and the computer-science education behind my work.

Frontend Developer — Intern

Aug 2024 – Nov 2024

Love In Property Pvt. Ltd.

  • —Built and shipped 20+ responsive UI modules using React.js and Tailwind CSS, maintaining consistent behavior and styling across devices and screen sizes.
  • —Integrated 10+ RESTful APIs to connect front-end components with backend services, enabling real-time data rendering and reducing perceived load times for end users.
ReactTailwind CSSJavaScript

Web Developer — Freelance

Nov 2024 – Dec 2024

M B Exim Pvt. Ltd. (Haniko)

  • —Built a static, production-ready e-commerce frontend for a honey-selling business using Next.js and Tailwind CSS.
  • —Integrated Brevo SMS API to automate real-time customer notifications and improve overall buyer engagement.
ReactNode.jsMongoDB

B.Tech, Computer Science Engineering

July 2022 – June 2026

Shri Shankaracharya Institute of Professional Management and Technology (SSIPMT)

  • —Class I (First Division) across all eight semesters, under CSVTU.
  • —Authored and developed the model for an IEEE conference paper during coursework — see Research.
04 — Toolkit

Working Toolkit

The tools I use across interface engineering, backend services, and AI-enabled products.

Frontend

Backend

Tools & Platforms

Selected technology

React.js

Frontend toolkit · 4 technologies

05 — Research

Applied AI Research

My IEEE OTCON 2026 work on Vision Transformer-based DeepFake detection, with model results and deployment details.

IEEE OTCON 2026

Deepfake Image Detection Using Vision Transformers (ViT) and Web Deployment

5th OPJU International Technology ConferenceAuthor · Web · Model Developer

A Vision Transformer-based approach for detecting manipulated face imagery, designed and trained as a full detection pipeline and deployed as a working web application rather than left as a research notebook.

0%

Accuracy on FaceForensics++

0%

Accuracy on DFDC

Vision Transformers (ViT)Deepfake DetectionWeb DeploymentComputer Vision
Read on IEEE Xplore
open to new roles & freelance work

Let’s build something
worth shipping.

Open to SDE-1 roles and interesting freelance work. Drop a message and I will get back to you.

> response time is usually under 24 hours
new-message