Connor Osborn

I worked with a designer to co-create a full-stack SaaS. This service handles authentication, payment, web crawling, and LLM integration and provides a consumable API.

Key technical challenges for this service include robustly extracting HTML from a website, identifying grammar mistakes in the text, handling false positives, and creating an efficient UI for selecting pages to spell check and processing many website pages in parallel.

I built a PDF compression tool that could run inside Figma (a browser environment). To run inside the browser environment, I had to cross-compile a modified version of the open-source PDF library QPDF (which backs pikepdf) to WASM. I made contributions to this project and added additional functionality for image compression.

The plugin has been used by 86,000 users and has about $1.5K MRR.

I designed and implemented a platform to allow scientists to upload analyses and users to run said analyses.

The platform allowed conforming docker images to be connected into pipelines. These pipelines were then versioned and could be passed files as inputs to produce outputs.

This service was implemented using a TypeScript front end, which subscribed to data changes via GraphQL. The back end for analyses was a Flask API, which launched docker containers in AWS ECS and managed files via AWS S3.

A service built with TypeScript, Flask, and GraphQL.

After working with this client to develop the first iteration of this project, I was hired to implement the customer-facing version. I worked with a design team and managed three other developers to complete this project. I also developed a docker-based development environment and hosted the service in AWS. I then used Terraform to manage a staging and production environment in AWS.