I focus on applied machine learning, computer vision, and learning-based systems. I enjoy turning ideas into working prototypes and documenting what I learn.
I am an engineering student exploring AI/ML and RL through hands-on builds. I prefer learning by shipping: small products, experiments, and technical guides.
A small set of projects that demonstrate how I build, document, and ship. Each card links to a live demo or repository.
A robust Streamlit application built to extract conversational insights through traditional exploratory data analysis, combining dynamic visualizations with the Gemini API for deep personality and behavior analysis.
A recommendation demo that suggests similar movies. Built to practice NLP-style feature extraction and user-facing ML delivery.
A collaborative web application built under hackathon constraints, designed to incentivize eco-friendly actions through a verified credit system and community-driven engagement.
Note:Currently backend is shutdown so website will not work
A lightweight practice website built to explore UI/UX and quick iteration ("vibe coding" exploration) with a clean interface.
A step-by-step technical guide published on GitHub and shared to the community (XDA). Focused on clarity and reproducibility.
Real-world solutions built for clients with practical business use cases.
Designed and delivered a custom mobile-first web application for Swadisht Mess, optimized specifically for Android users to strengthen digital presence, improve customer interaction, enable seamless menu access, and streamline daily service workflow. Developed as a real client project with production-focused thinking, practical usability, and business impact in mind.
I write to clarify concepts and document progress. This is also where I share broader research summaries.
A high-level timeline of NLP milestones, from early statistical methods to modern transformer-era systems.
A first-semester research experience focused on abuse detection using a hybrid heuristic–machine learning approach, later presented as a poster.
Submitted under the guidance of Sundeep Deulkar Sir and presented as a poster at CTFC 2025. The work explores slang normalization, TF-IDF features, and machine learning for abuse detection.
I keep the stack lean and practical. I focus on fundamentals, reproducible experiments, and clear demos.
I am particularly interested in reinforcement learning and agentic systems - how agents learn from interaction, how we evaluate them, and how we build reliable behavior under constraints.
RL is a practical way to study learning-by-doing. It connects perception, decision-making, and feedback loops, which makes it a strong foundation for building intelligent systems beyond static prediction.