Product Builds & Hackathons

Illustration of AID's merchant catalog ingestion, vector retrieval, and agent recommendation flow

AID — Agentic Advertising Platform

September 2025 - October 2025

Built a backend to explore how agentic commerce and advertising could work, turning merchant product catalogs into vector-searchable listings and serving hybrid recommendations to authenticated AI agents.

Designed & implemented backend architecture using FastAPI and OpenSearch to ingest structured merchant product catalogs and serve recommendations to authenticated AI agents
Engineered product ingestion pipeline that normalized product metadata, generated contextual embeddings via OpenAI’s Embeddings API, and indexed listings as vector-plus-metadata documents for semantic retrieval
Built hybrid recommendation system combining OpenSearch k-NN vector search with structured filters across price, brand, category, and attributes, then reranked results using weighted semantic similarity and product completeness scores
Implemented JWT-based affiliate tracking, impression logging, click redirects, and idempotent conversion reporting to measure downstream recommendation performance

PythonFastAPIOpenSearchVector SearchJWT

Project Demo

Screenshot from NVIDIA AI lecture assistant demo

AI Lecture Note-Taking Assistant — NVIDIA GTC 2025 Golden Ticket Submission

January 2025

Backend for an AI-powered lecture companion that turns raw audio into structured Google Docs notes using speech-to-text + LLM pipelines.

Designed & implemented backend system that converts lecture audio into structured, comprehensive notes using LLM pipelines
Engineered end-to-end workflow integrating speech-to-text, semantic summarization, and hierarchical note structuring synced with Google Docs
Built scalable API-based architecture to process long-form lectures while preserving key concepts, examples, and context
Positioned tool for dual-use by students and professors to improve accessibility, retention, and material distribution

PythonLLMSpeech-to-TextAPIs

Project Demo

Post Filtering Extension — 1st Place Winner, xAI Hackathon (Replit Award)

October 2024

Chrome extension that filters engagement farmers, self-promoters, and bots on X using Grok 2 / xAI APIs.

Academic Support Platform — Google Gemini API Developer Competition Submission

August 2024

Backend for a Gemini-powered academic assistant that simplifies registration and real-time support.

Engineered backend architecture of web application designed to enhance academic life by simplifying course registration & providing real-time academic support using Python, Flask, & effective prompt engineering for Google’s Gemini API
Utilized Google’s Gemini API to develop AI-driven Virtual Tutor & Academic Advisor tools for personalized student assistance
Integrated Firebase for secure data storage & efficient management of chat histories, ensuring reliable app performance

PythonFlaskGeminiFirebase

Project Demo

Career Fair Data Analysis Automation

April 2023

Automated the Santa Clara career center’s fair analytics workflow so peer advisors could see student + employer trends without spending a week inside Excel.

Health Match — Hack4Humanity Submission

February 2023

Hack4Humanity build that pairs symptoms with AI-powered specialist recommendations using Flask + OpenAI’s DaVinci 3 API.

Tripfol.io — 1st Place Winner, INRIX Hackathon @ Santa Clara University

November 2022

INRIX hackathon winner that builds optimized day-trip itineraries with a custom routing algorithm over INRIX APIs.

Designed & implemented application backend using Python, Flask, & INRIX Routing API to develop optimized routing algorithm for users that generate an itinerary based on hotspots in a geographic location
Identified & implemented relevant graph theory to develop optimal routing algorithm in under 24 hours
Implemented Flask API to return line strings of coordinates to enable front-end map routing on Google Maps plugin

PythonFlaskINRIX APIsGraph Theory

View Hackathon Demo

Project Pineapple — 2nd Place Winner, Hack4Humanity

February 2022

Safety-focused hack that schedules check-ins for nights out using a Python/Flask backend and SQLite storage.

Designed application backend using Python, Flask, & SQLite that uses scheduled check-ins & user verification to allow a safe night out to reduce the likelihood of sexual assaults
Implemented an SQLite database using Python programming to store & maintain data of users effectively & systematically
Developed an API using Flask to fetch data from database & link to the front-end software for an efficient full-stack software

PythonFlaskSQLiteSafety

View Hackathon Demo