Rutwik // Portfolio
ENGINEER
From Concept to Deployment.!
From Concept to Deployment.!
I don't just write code; I engineer systems. As the Vice Chair of the IEEE CS Student Branch (25-26) and a Full-Stack AI Developer, I thrive at the intersection of Research, Product, and Architecture.
Whether it's architecting serverless fintech solutions for non-profits, optimizing NLP pipelines for endangered languages, or leading technical teams — my goal is to build software that is scalable, meaningful, and production-ready.
I specialize in taking raw concepts — be it a research paper or a business need — and deploying them as robust, user-centric applications.
A pioneering NLP initiative to preserve the endangered Gondi language. Co-authored the core lexical database and processing models, culminating in a peer-reviewed IEEE publication.
Developed custom tokenization and lemma processing pipelines specifically optimized for sparse datasets where standard pre-trained models fail.
Architected a complex SQL schema to map intricate linguistic relationships (synsets, hypernyms) while maintaining sub-100ms query performance.
The methodology was validated by the research community and published in the IEEE ICAIQSA Conference 2024.
A full-stack AI study platform solving information overload. Currently scaling with 350+ users and 580+ decks, generating interactive flashcards instantly from unstructured data.
Engineered a Retrieval-Augmented Generation pipeline using LLMs to contextually extract Q&A pairs from dense academic texts with high accuracy.
Built a robust parsing layer capable of normalizing messy inputs from PDFs, DOCX files, and raw web articles into structured JSON for AI processing.
Optimized database indexing and API response times to handle concurrent generation requests without latency spikes.
A full-stack solution for a local ISP. The site features a custom Flask-based admin dashboard for managing users and plans, and a custom payment gateway integration for payments, increasing their online sales conversion.
A high-performance digital infrastructure handling thousands of monthly visitors. It features a proprietary FinTech layer engineered to maximize transaction success rates while maintaining zero operational overhead.
Developed a custom payment protocol that intelligently bridges web browsers with native UPI apps, significantly reducing transaction drop-offs compared to standard gateways.
Architected a dual-stage verification engine that issues instant provisional acknowledgments to users while asynchronously queuing financial data for official audit and tax receipt generation.
Engineered a fully decoupled architecture where complex computations (astronomy algorithms & document generation) are offloaded to the client-side, resulting in near-zero server costs at scale.
A high-performance 3D endless runner developed in Unity. Engineered for smooth physics and deployed to itch.io within a rapid one-week prototyping cycle.
Implemented efficient collision detection and object pooling to maintain stable 60FPS even in WebGL browser environments.
Designed algorithms to spawn obstacles dynamically, ensuring unique gameplay loops for every session.
Optimized assets and lighting baking to ensure the game loads instantly on low-end devices without requiring downloads.
As a co-author, I developed the data-gathering processes and NLP models for building a lexical database for the endangered Gondi language. This project contributes to linguistic preservation and was formally published at the 2024 International Conference on Artificial Intelligence and Quantum Computation-Based Sensor Application (ICAIQSA).
