





AI Engineer • Physics Fanatic • Linux Lunatic
I like to build things and break things and build again, but even better this time. I draw my happiness from the moments when I make things work
My journey into artificial intelligence started at Delhi Technological University, where I studied Engineering Physics with a minor in Mathematics and Computing. There I gained the rigorous mathematical foundation behind machine learning and AI.
But coursework was only the starting point. The real progress came from hands-on building.
May 2025 - Present • Remote
Today, I'm navigating the cutting edge of AI at C Simplify IT, where every day is about pushing boundaries. I work with the Hugging Face ecosystem, building AI applications that feel like magic but are grounded in rigorous engineering.
Fine-tuning large language models like Qwen3 using LoRA has taught me that AI isn't just about raw computation—it's about personality, nuance, and behavior. I've created chatbots that don't just respond, but converse with distinct intonations and character.
Working with FastAPI and MongoDB, I handle the invisible but crucial work: data pipelines, infrastructure, and the engineering that makes AI products actually work in the real world. I even forked the Web Thinker open-source project, adapting it to breathe life into new applications.
Each project is a star in my constellation—distinct, purposeful, and part of a larger pattern.
What if an AI could understand the stock market like a seasoned analyst? I built an agent that researches NSE stocks using ICICI Breeze and Zerodha Kite APIs. It doesn't just crunch numbers—it listens. Using Whisper for speech recognition, you can literally talk to it. Built with Smolagents and OpenAI's agentic frameworks, it features a Brave Search tool for real-time market intelligence. Gradio provides the interface where human intuition meets machine precision.
Smolagents • OpenAI Agents • Gradio • Whisper • ICICI Breeze API • Zerodha Kite API • Brave API • BeautifulSoup
Information trapped in blog posts is information lost. I created a RAG pipeline for the 'Please Explain Me' blog that makes knowledge searchable and conversational. Using Gemini's text embedding API with MongoDB Vector Search, I turned static articles into a queryable knowledge base. The bot scrapes Gartner's business trends with Selenium, chunks the data intelligently, and responds using gemini-2.0-flash. It's like giving a blog a brain.
RAG Pipeline • Gemini API • MongoDB Vector Search • Selenium • gemini-2.0-flash
Being a car enthusiast, I always felt the need to have a collection of them, but since its not practical to buy all these gas-guzzling beasts, I present a way to add these custom models yourself and marvel at all the other car models at the Coldwheels platform. The website is powered by Vue.js, and the backend server is made using Go programming language and Gin Web Server, containerized using Docker (cyt0/gowheels on Docker Hub)
Vue.js • Go • Gin • GORM • PostgreSQL • Docker • Pinia State Management • Daisy UI
Before diving deep into AI, I needed to master the fundamentals of building robust Full Stack applications. Outstagram is a complete full stack application—not a toy project, but production-ready infrastructure. Built with FastAPI and Vite/React/TS, it handles everything: user authentication with argon2-cffi, session management, media uploads, likes, comments, friendships, and follow requests. Powered by SQLAlchemy ORM and PostgreSQL, containerized with Docker (cyt0/outstagram on Docker Hub), it features pagination and feed algorithms. This taught me that AI without solid engineering is just a demo.
FastAPI • SQLAlchemy • React.js • Typescript • Bun • PostgreSQL • Docker • Session Management
Web Dev Lead • 2023-2024
The best way to learn is to teach. As Web Dev Lead, I shared knowledge with over 100 students through offline events and YouTube videos, covering web development, networks, Django, and best practices. Every question they asked made me dig deeper into my own understanding.
Volunteer • June 2024
Technology is powerful, but purpose gives it meaning. I volunteered with NSSS Foundation NGO, building their web presence and coordinating their Yoga Day 2024 event. It reminded me that behind every line of code should be an intention to make the world a little better.
Like an astronaut needs their equipment, I've mastered the tools that let me explore the AI frontier.
Python • TypeScript • Go • C++ • Bash • MATLAB • SPSS • x86 Assembly
FastAPI • Django • SciPy • NumPy • SQL • MongoDB • Nginx • Machine Learning • Computer Vision
Vim • Docker • Git • Cron • AWS S3 • Linux
An exploration into the physics of projectile hitting a target, often useful for predicting ballistics
A deep dive into maths of finding the edges using Operator, with a follow-up blog featuring a Jupyter Notebook
A look into why does the most used optimization algorithm in Machine Learning work?