I'm Pravin — a Full-Stack Developer and student researcher at R.M.K Engineering College. I ship real products, from a compliance-intelligence dashboard to a note-sharing platform my classmates use every day, and co-authored IRRIS, an IEEE-published sorting algorithm.
Currently pursuing a B.Tech in Information Technology, I spend most of my time at the intersection of design sensibility and engineering precision. I build web applications that are fast, accessible, and considered — from the first pixel to the last API call.
I'm not just a frontend developer who dabbles in backend. I'm deliberately building depth across the full stack, with a growing focus on AI integration and systems that scale. My goal is simple: write software that matters.
Frontend
HTML5
CSS3
JavaScript
React.js
Next.js
Tailwind CSS
Backend & Data
Python
Flask
Node / Express
MongoDB
MySQL
Java
Tools & Design
Git
GitHub
Figma
Canva
Programming
C
C++
A campus note-sharing platform built for students of R.M.K Engineering College.
A regulatory change analyzer: what changed in the law, and which of our policies does it break?
A full-stack internship & job platform — an Internshala-style experience end to end.
An ML-powered Stock & ETF signal engine that answers one question: BUY, HOLD or SELL?
/stocks, /predict/<ticker>) with confidence scores.
A financial transaction management and analytics dashboard for banking data.
Real data is rarely random: it's full of stretches that are already ascending or descending. IRRIS is an adaptive, stable sort that detects both kinds of run in a single pass, reverses the descending ones, and merges the runs pairwise in balanced rounds. A hybrid_merge() step checks run boundaries in O(1) and simply concatenates runs that don't overlap, skipping merge work entirely.
The paper proves correctness, termination and stability, and benchmarks IRRIS against CPython's Timsort. IRRIS is up to 11× faster on sorted, reverse-sorted and partially reversed input, while Timsort stays stronger on random data.
| Input | IRRIS | Timsort | Result |
|---|---|---|---|
| Sorted | 0.36 ms | 3.89 ms | IRRIS 10.7× |
| Reverse-sorted | 0.38 ms | 4.21 ms | IRRIS 11.0× |
| Partially reversed | 0.62 ms | 4.00 ms | IRRIS 6.5× |
| Nearly sorted | 32.45 ms | 6.50 ms | Timsort 5.0× |
| Random | 84.15 ms | 13.35 ms | Timsort 6.3× |
A two-stage computer-vision pipeline that first locates and crops the diseased region of a tea leaf, enhances it, and then classifies it with an ensemble of modern vision transformers and CNNs, so the model learns from the lesion rather than the background.
A multi-model machine-learning framework that turns a live ball-by-ball scorecard into one ranked, explainable tactical recommendation for a cricket coach during the strategic time-out.
Built a full-stack internship and job listing platform (InternArea) with Next.js, Node.js/Express, and MongoDB. Implemented secure Google Auth via Firebase, Razorpay subscriptions, an Admin Dashboard, Redux Toolkit, and handled end-to-end deployment.
Designed and developed FinFlow, a financial dashboard, in a team environment. Deepened API integration skills and problem-solving capability while focusing on clean, production-ready UI.
Explored backend technologies and AI model integration — training and serving ML models behind a Flask API for stock-signal generation. Worked on real-world problem statements and delivered scalable solutions.
My inbox is always open. Whether you have a project to discuss, a role to fill, or just want to say hello — I'd love to hear from you.