2023-present: Using ML for climate and weather modeling

2018-2023: Using ML to discover novel imaging biomarkers (e.g. https://www.nature.com/articles/s41551-022-00867-5).

Orbital Insight is developing technology to analyze imagery of the Earth (from satellites, UAVs, etc) at scale using state of the art computer vision and machine learning techniques.

* First full-time computer vision engineer at the company, and currently an eng. lead. Spearheaded efforts to grow the team (interview pipeline, sourcing, etc.) up to 7 people.

* Built infrastructure to support training and deploying computer vision models: tools and API for managing dozens of datasets, common API and training tools for objection detection and semantic/instance segmentation, data modeling to keep track of models and input requirements for each (resolution, spectral bands, etc.)

* Developed computer vision algorithms for a number of products: detecting various types of vehicles, segmenting buildings, measuring amount of crude oil in storage tanks, estimating income levels, etc. Most of these systems were built on top of deep learning / convolutional neural nets. Responsible for developing these systems end-to-end: dataset collection, prototyping / evaluating different algorithms, productionizing them to run at scale.

Here's a talk I gave at AAAI'18: https://www.youtube.com/watch?v=HJlYpY7CZqA

Software engineer on the "photo science"​ team (sub­team within photos team that focuses on leveraging computer vision & machine learning to help create a magical photos experience for Dropbox users).

* Worked on web front­end and back­end parts of the stack to help launch web photo product (the entire photos team was ~5 people at this time).

* Built and launched the first computer vision system at Dropbox (see http://www.wired.com/2014/04/3-ingenious-design-details-in-carousel-dropboxs-new-photo-app/ for some hints of what it involved): collected an annotated dataset, researched and implemented a set of algorithms from the literature, benchmarked against open source and commercial systems, optimized implementation via SSE4, and productionized to run at Dropbox scale.

* Spearheaded efforts to grow the computer vision / machine learning team: developed interview pipeline, sourced and evaluated candidates and acquisition opportunities. Team went from two engineers in 2012 to 10 engineers in 2015 when I left.

Co-founded a company that built a prototype product for automated scientific image analysis (microscopy, satellite imagery, etc) via combination of computer vision and crowdsourcing. Took this concept through the customer discovery process and eventually pivoted into consumer photo auto-organization.

* Went through the NSF ICorps program (via Caltech) and the Summer @ Highland incubator.

* Applied for and received an NSF SBIR Phase 1 grant.

* Company was acquired by Dropbox in September 2012.

Worked on a project related to next-gen sequencing.