Projects
I like solving problems that pull at my curiosity, and I usually do it through code or words. Building things lets me explore new ideas, tinker with systems, and understand the world a little better while also sharing my view of it. You’ll see traces of the things I love scattered across these projects: language, distributed systems, machine learning, football, and art. If something catches your attention, tell me, I’d love to hear about it!
Salient Entity Linking 2025
A training-free workflow to idenity and link salient entities in web articles.
sel is an experimental system that detects the important entities in a webpage (primarily news articles) and links them to canonical references. It’s designed to be fast, accurate, and easy to integrate into indexing pipelines.
This is an ongoing independent research project we’re developing in collaboration with Apple. The direction is still evolving, and it may look completely different a few months from now but that’s the fun part :)
Links: GitHub
Artwork Search 2025
Searching for artwork using themes and keywords, built for writers and creatives.
artwork-search that helps writers (including myself) find artwork based on themes, moods, and textual prompts. It’s meant to be a lightweight discovery tool; a way to quickly gather visual references when I’m working on essays or stories.
Links: WIP !!!
Onestop football 2025
A platform for player similarity search, scouting, and quick stat discovery.
onestop-football is a football analysis tool with an interactive formation visualizer. Click on a player to find similar profiles, check market value, explore stats, or browse average positions and default team shapes. Much more to come!
Links: WIP !!!
Tactical Touchline 2025
My home for all things football analytics.
Reflective Prompting for RAG 2024
Course project for CS685 exploring reflective prompting in RAG
reflective-prompting explores a cost-efficient alternative to fine-tuning by letting the model iteratively refine its own retrieval process. We built a custom Piazza QA dataset from our CS685 course materials and compared vanilla RAG, Self-RAG, and our self-refine approach. Our method achieved the best factual accuracy, showing how multi-step reflection reduces hallucinations without training overhead.
Links: GitHub
Swipe My Thesis 2024
A Tinder-like research paper discovery app built at HackUMass.
swipe-my-thesis is a gamified way to discover research papers from arXiv. Users swipe through summaries, save what interests them, and even match with others who share similar academic tastes. We built it during HackUMass 2024, and while the deployment is currently down, it’ll be up again soon.
Links: GitHub
ASAG - Automatic Short Answer Grading 2023
An Automatic Short Answer Grading system developed at CMInDS, IIT Bombay.
asag uses linguistic and embedding-based features to assess student answers relative to model responses. I built this during my research internship at C-MInDS, experimenting with multiple similarity metrics and regression models to outperform prior results on the Mohler dataset.
Links: GitHub
Amherst PVTA Transit Streaming Service 2023
A real-time tracking service for PVTA buses in Amherst, Massachusetts.
pvta-transit-streaming delivers live updates on bus locations, ETAs, and routes for students in Amherst. Built as part of CS532 (Systems for Data Science), it introduced me to event-driven architectures and gave me my first real exposure to Kafka.
Links: GitHub
Explainable Plant Species Identification 2022
Identifying plant species from leaf images with explainable AI.
xai-vnplant200 is a deep-learning classifier augmented with LIME to highlight the regions influencing predictions. It was my first exploration into explainability.
Links: GitHub
FoodChain 2020
A platform to redistribute excess food and reduce waste.
foodchain connects people or businesses with surplus food to those who need it. Users can list available items, browse nearby listings, and coordinate pickup. This was also my first hackathon :) We got into Top-25!
Links: GitHub