Product UI
Interfaces that survive contact with real users - flows, states, and the fiddly bits nobody specs.
Available for freelance & internships - 2026
I buildweb apps
Software engineer and UI/UX designer from India. I design first, then build - usually in the same sitting. Right now: finishing my B.Tech in Computer Science (AI & ML) and taking on freelance work.

I'm a B.Tech Computer Science student with an AI & ML concentration at Dronacharya College of Engineering. This summer I interned at Surakshit City Pvt Ltd, building responsive frontends and product UI for a smart-city platform.
Before that - three years of freelancing for clients worldwide: marketing sites, web apps, design systems and the occasional very stubborn CSS bug. I like work that looks considered, reads clearly, and survives being handed to someone else.
Interfaces that survive contact with real users - flows, states, and the fiddly bits nobody specs.
Tokens, components and docs so the tenth screen looks like the first one, on purpose.
PyTorch experiments and Hugging Face models wired into products that actually ship.
Surakshit City Pvt Ltd
Two months on a smart-city platform - shipping responsive frontends and product UI end to end, and learning the backend half of the craft.
Self-employed
Interfaces, marketing sites, design systems, deployments and web applications for clients worldwide.
Dronacharya College of Engineering
Computer science with an AI/ML concentration: linear algebra, data structures, and the machine learning that comes after them.
5 repositoriessynced Oct 20, 2026
LawgicAI is an AI-powered legal assistant focused on Indian law. It provides instant, context-aware answers to legal questions, document search, and multilingual support, all with a modern web interface.
Building a 15M parameter biomedical LLM from scratch - end-to-end pipeline including tokenization, pretraining, LoRA fine-tuning, quantization, and deployment.
An AI-powered web application that automates financial document analysis with real-time updates using local LLMs via LM Studio.
The website is a browser-based mini-games collection platform. It presents a set of simple, interactive games that users can play directly without downloading or signing in. The focus is on quick engagement and lightweight gameplay.
A web-based application that uses machine learning to classify news articles as Fake or Real. Features include user authentication, prediction history, comments section, and a responsive design. Built with Flask, MySQL, and a Passive Aggressive Classifier trained on TF-IDF Vectorizer.
Everything is open source - github/rohansingh-dev
A toolbox, not a personality - the interesting part is what gets built with it.
Got an idea, a product, or a stubborn problem?
Say the word.
I answer every message personally - collaboration requests, freelance jobs, internships, or just a hello.