//
First-year B.S. student at IIT Jodhpur building real-world AI — financial intelligence, trading bots, RL agents, and Pinterest AI. Every project is open-source on GitHub.
The best models aren't just accurate — they're understandable, fair, and built to solve real problems for real people. I build AI not to impress, but to improve.
Every project starts with a real problem — loan approval bias, trading noise, RL instability, even Pinterest automation. If it doesn't solve something, it doesn't ship.
All 6 projects live publicly on GitHub — fully documented. Building in the open is how I get honest feedback and contribute to the community.
Finance × NLP, RL × game theory, automation × vision — I live at intersections where the most interesting ML problems hide.
First year at IIT Jodhpur. Already shipped 1 internship, 6 open-source projects, and 1 hackathon entry. The best work is still ahead.
Joined one of India's premier technical institutes. Maintaining 8.7 CGPA while running live ML projects and competing in international hackathons. Coursework: ML, Deep Learning, NLP, Data Engineering, Computational Statistics.
Built a production sales forecasting dashboard (10K+ data points, −40% reporting time) and deployed an NLP chatbot with ~85% intent accuracy using GPT + Dialogflow. Resolved 3 critical business bottlenecks.
Competed in Meta AI-backed RL hackathon. Implemented deep RL agents across 5 OpenAI Gym environments using PPO — achieving policy convergence within 500 episodes. Code fully open-sourced.
Represented IIT Jodhpur at IIT Gandhinagar's flagship festival. Built cross-IIT networks and drove student engagement across top engineering campuses.
An intelligent Pinterest automation agent built with HTML + AI. Automates pin browsing, content curation, and interaction workflows — combining browser automation with AI decision-making for smart social media management.
Real-time AI trading assistant for Indian markets. Fuses live NSE/BSE price data, RSI/MACD technical indicators, and GPT-powered news sentiment analysis into clear buy/sell signals with explainable reasoning.
Competed in Meta AI-backed RL hackathon at Scalar School. Implemented deep RL agents using PPO across 5 OpenAI Gym environments, achieving consistent policy convergence. One of the most challenging builds to date.
Intelligent ML loan approval classifier supporting accurate, unbiased financial decisions. XGBoost & Logistic Regression trained on 10+ financial/demographic features with cross-validated pipeline engineering.
Comprehensive data analysis capstone covering the full data science pipeline — from raw data ingestion with Python & Pandas, spreadsheet processing, Orange ML toolkit exploration, to MySQL-backed reporting and insights generation.
A Python chatbot built to deliver live hackathon news and updates. Fetches the latest hackathon listings, deadlines, and registration info — keeping student communities informed without manual monitoring.
Building attention mechanisms and multi-head attention from scratch in PyTorch — going beyond just using HuggingFace APIs.
Going deeper into policy gradient methods beyond PPO. Exploring real-world decision applications outside classic game environments.
Productionizing models with FastAPI and Docker — bridging the gap between Jupyter notebooks and real deployed applications.
Selected to compete in OpenEnv RL Hackathon backed by Meta AI — implementing deep RL agents in competitive gym environments at Scalar School.
Represented IIT Jodhpur at IIT Gandhinagar's flagship technical festival, driving cross-campus student network expansion.
Maintained strong academic performance in a highly competitive programme while simultaneously shipping 6 open-source projects and 1 internship.
From a 91%-accurate loan classifier and real-time trading AI to a Pinterest agent and hackathon chatbot — all public, all documented.
Open to ML research internships, AI engineering roles, and collaborative projects that push the frontier. Reach out — let's build something that matters.