This is me — in code.
I’m always tinkering with new ideas, building apps, and learning by doing.
I enjoy making tech that feels simple and useful — and yeah, caffeine helps. When I'm not coding,
I made karan.solutions as a space to share what I’m working on, what I’m learning, and maybe even a few decent ideas along the way.
FitOrbit – Adaptive Fitness App
FitOrbit is a cross-platform mobile fitness app built with Flutter that personalizes and adapts calorie burn goals using AI. Users input their body metrics, and the app calculates a plan to meet fitness targets within a set timeframe.
If a goal is missed, the app intelligently adjusts the remaining targets using a lightweight LLM model from Hugging Face.
Tech Stack: Flutter, Python, Hugging Face API
Vision-Aware AI Parking Assistant for EVs
Built a smart AI assistant that helps EV drivers park more responsibly using computer vision and a language model. The system uses YOLOv8 to detect sidewalks, crosswalks, and no-parking zones, then explains—in simple, natural language—why a spot might not be ideal.
It’s designed to promote better urban parking habits and ensure accessibility for everyone. Combined edge-based vision with an open-source LLM for real-time feedback and reasoning.
Lazy Polona - The chatbot with better things to do.
Lazy Polona is a mood-based AI chatbot with attitude — sometimes helpful, mostly toxic.
Built using open-source LLMs like OpenChat, it replies with sarcasm, roasts, and chaotic humor.
Hosted on an EC2 instance using Ollama for fully free, offline, and customizable AI interaction.
