Building a socratic reading companion app!

Owen Hochwald  • Peerlist

Built RecMind, a distributed RAG knowledge-based system.

RecMind |  Knowledge Base Recommender System

A high-performance, terminal-based HTTP client built for developers who live in the command line. Supports up to 100k concurrent requests/ sec as a Postman and K6 replacement.

Volt

Through analysis of recipe popularity data and the implementation of supervised machine learning techniques, I successfully recommended high-traffic recipes to a stakeholder. By working through the data science life cycle, I trained reliable machine learning models to improve user engagement and increase in website traffic for the company.

Report: Predicting Popular Recipes for Websit

UBCStudyAssist is an AI-powered web application designed to empower students at the University of British Columbia (UBC) and beyond by providing personalized study strategies tailored to their individual needs and academic goals. The app leverages the latest advancements in Large Language Models (LLMs) to create an intelligent chatbot that helps students develop effective and customized study plans.

Study Assist UBC

A WIP, free-forever time tracking web application for business and personal use to enhance your productivity!

Aftrbrnr Time Tracking SaaS

A friendly cat-themed chrome extension  to remind the user to maintain good pawsture while working! Features: • Customizable reminder intervals • Daily posture tracking - Watch your progress and build healthy habits • Clean, modern interface - Easy to use and easy on the eyes • Multiple notification styles - Choose between system alerts or popup windows • Minimal permissions required - We respect your privacy!

Pawsture Pal

Built AdmitifyAI, an AI-powered college admissions platform that helps students find their best-fit schools and improve their chances of acceptance. I led full-stack development and core AI integrations, including admissions probability models based on Common App-style data and personalized college matching algorithms.

AdmitifyAI | Your Perfect College Match

Created a GPT-3-like transformer model trained on the first 1,000,000 characters of Shakespeare's work using an architecture based on the research paper "Attention Is All You Need". I started on a simple Bigram model to measure the output. From here I iterated, adding additional features like multi-headed self attention, and residual connections, to imitate the style of the input data.