





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
I'm currently working as an Software Engineer at C Simplify IT, where I usually tinker with cutting edge voice AI, STT, TTS, OCR, and Generative AI. 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 using Runpod and Docker, and the engineering that makes AI products actually work in the real world. I also contribute to open-source projects, which are used in our services, thus giving back to the community.
Each project is a star in my constellation—distinct, purposeful, and part of a larger pattern.
Pariksha is a exam practice platform designed for scientific exams. It features a exam engine grounded with scientific papers and books embedded into a Vector Database, coupled with an automated AI pipeline for question generation and question bank management. It features Topic wise custom exams, Negative Marking section wise systems, MSQs, NATs, and a live scoring sidebar that tracks potential max/min scores via real-time session telemetry
LlamaIndex • ChromaDB • RAG • Gemini API • Cloudflare R2 • React • Tailwind CSS
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
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?