(02) Work

Selected
work.

Six projects, each shipped end to end: models, contracts, interfaces and the glue in between.

(06)Scroll to browse →

01 / 06

Final-year projectJul 2026 – present

STOCK AI

AI-driven stock forecasting & an avatar-presented market newsroom.

The platform predicts prices with statistical, machine-learning and deep-learning models and analyzes real-time financial news alongside them.

ARIMA/SARIMA baselines compared against XGBoost and LSTM forecasts
FinBERT sentiment scoring on live headlines
AI-avatar news briefings, technical indicators and TradingView charts

Python
TensorFlow
scikit-learn
FinBERT
Yahoo Finance API
React
TypeScript
Node.js
Tailwind

Live site ↗GitHub ↗

02 / 06

Open source · Web3Mar 2025

NFT Vault

Decentralized asset management for creators.

An open-source platform where artists mint, manage and monetize NFTs, with royalties enforced on-chain and media authenticity backed by decentralized storage.

Multi-chain deployment on Ethereum and Polygon
Lazy minting, auctions and creator dashboards
IPFS-backed provenance tracking for every asset

Solidity
Ethereum
Polygon
Web3.js
IPFS
Chainlink
Hardhat
React

GitHub ↗

03 / 06

DeFi protocolDec 2024

DeFi Vault

Lend, borrow and earn with no intermediaries.

A decentralized lending and yield platform where smart contracts automate collateral, interest and liquidity, with real-time analytics on top.

Collateralized loans with dynamic interest rates
Yield-farming pools for liquidity providers
Chainlink oracle price feeds to keep liquidity healthy and limit risk

Solidity
Ethereum
Chainlink
Web3.js
React
Hardhat
IPFS

GitHub ↗

04 / 06

Smart India HackathonSep 2024

Cyber Trigger

An ML-powered digital forensics assistant.

Built for SIH, it speeds up digital forensic investigations by automating how incident data is collected, analyzed and reported, so responders spend less time and make fewer mistakes.

Evidence import from disk images (libewf, pytsk3)
AI/ML anomaly detection across collected artifacts
Investigator-friendly interface with full reporting

Python
pytsk3
libewf
Wireshark
Solidity
Hardhat

GitHub ↗

05 / 06

Machine learningJan 2024

Movie Recommender

Content-based movie suggestions from a single favourite title.

A recommendation system that suggests 30 similar films from a dataset of 4,803 movies, using each film’s genres, keywords, tagline, cast and director.

TF-IDF vectors built from five combined text features
Cosine similarity across a 4,803 × 4,803 matrix to rank matches
Fuzzy title matching with difflib, plus exploratory charts of ratings and popularity

Python
pandas
NumPy
scikit-learn
seaborn
Google Colab

Open in Colab ↗GitHub ↗

Restaurant website hero: "Welcome To Our India Restaurant" next to a sushi photo

06 / 06

Front-end2023

Restaurant Website

A one-page, animated website for an Indian restaurant.

A single-page restaurant site with a filterable menu, gallery, table booking hours, chef and testimonial sliders, FAQs, a blog and a newsletter footer.

Menu filtered by breakfast, lunch and dinner with MixItUp
Swiper sliders for chefs and reviews, and a Fancybox gallery
Smooth scrolling and parallax effects with GSAP ScrollTrigger

HTML
CSS
JavaScript
Bootstrap
jQuery
GSAP
Swiper
Netlify

Live site ↗GitHub ↗

Keep scrolling

How I work

Research. Build. Ship.

Research the problem

Start from the papers and the data. For STOCK AI that meant benchmarking classical ARIMA against ML and deep-learning models before writing any UI.

Build the core

Get the hard part right first, whether that’s a forecasting pipeline, a lending contract or an evidence parser, then wrap it in clean APIs.

Ship & polish

Deploy early (Vercel, GCP, testnets), get feedback and keep iterating on the details people notice.

Next: the toolboxStack

Selectedwork.

STOCK AI

NFT Vault

DeFi Vault

Cyber Trigger

Movie Recommender

Restaurant Website

Research. Build. Ship.

Research the problem

Build the core

Ship & polish

Selected
work.