Perception and other software for Online Mapping team

As the one-man Style AI team at Modsy, I work on projects applying machine learning and intelligent algorithms to improve Modsy's internal styling tools and customer marketplace experience. This includes:

• Building algorithms quantifying how stylistically cohesive interior design products are and to select optimally cohesive populations of products subject to specific constraints.

• Using concepts from the recommender systems literature to improve our product search implementations

• Using self-supervised and fully supervised representation learning techniques to implement visual search

• Working on everything from ML research and training to ML Ops to integrating my work into our frontend services - my stack includes PyTorch, Sagemaker, S3, Lambda, ECS, MySQL, Elasticsearch, backend work in Django, and frontend work in React

As a member of Modsy's room-building research team, I built computer vision and machine learning tools to help automate the construction of 3D models of interior spaces from captured images. This includes:

• Conducting literature reviews and adapting convolutional neural networks for specialized segmentation tasks

• Building a synthetic data generation pipeline to create novel rendered views of existing 3D room models and accurate ground truth data.

• Designing and building a visual SLAM implementation to utilize captured pose data for better accuracy than available open source solutions.

• Other work includes traditional computer vision tasks, optimization problems, backend pipeline code in Python, frontend work in JS + React, and experimental features fusing classification models with 3D spatial reasoning algorithms

• Co-author on Modsy's pending patent for our room-building pipeline

Built visualizations of company progress metrics in Tableau.

• Developed dynamic ingestion counter for display in company lobby.

• Built internal Tableau proof-of-concept for democratizing data pulls.

• Worked extensively with Amazon Redshift, Postgresql, Python, Javascript.

Designed and led workouts for multiple age groups, taught swim lessons, managed junior coaches, optimized meet entries, ran team social events over numerous years.

Lead Assistant Coach, Summer 2016

Senior Coach, Head of 11-12 age group, Summer 2014

Senior Coach, Co-head of 6 & under, 7-8, 9-10 age group, Summer 2013

Junior coach for Palo Alto Stanford Aquatics club swim team, Fall 2012-Summer 2013

Junior coach, 2009-2012 summers