Austin, Texas, United States
I am an experienced back-end software generalist, with a low-level embedded systems background, currently interested in distributed systems in cloud environments. My background in computing and programming started in college where I studied the architecture of computers (CPU and memory design) and embedded systems. I was skilled at C, had an understanding of hardware design, and even worked in a flash memory fabrication facility. Over the last 10 years I've worked in a wide variety of companies in the software industry. I have a good understanding of the big picture of software architecture, all the way from a micro-service running in the cloud, down to the bare metal. I am now focused on building distributed software systems in modern cloud environments. I like to accumulate a broad range of skills about how cloud computing products are built, opposed to specializing in one area. My long term goals are to continue to increase my skills in the end-to-end process of SaaS product production, as well as the technical aspects of distributed systems. I can program in, or quickly learn, any language, however my favorite languages are C++, Go and Python. I have a total of 12 years of experience programming. I'm comfortable working in small software teams in start-ups, or at large mature organizations.
Working inside the query engine. My team specifically is focused on enabling / improving features related to full-text search and vector (semantic) search. Worked as the engineering lead for the release of MongoDB's initial Hybrid Search offering, leading 3 other engineers. Me and the team released $rankFusion (https://www.mongodb.com/docs/rapid/reference/operator/aggregation/rankFusion/) in 8.1, and $scoreFusion (https://www.mongodb.com/docs/manual/reference/operator/aggregation/scoreFusion/) in 8.2.
Nightfall is a SaaS start-up (backed by Venrock, Bain, PearVC, and others) that provides a Data Loss Prevention (DLP) solution that alerts its clients of sensitive information (i.e SSNs/CCs) that gets exposed inside their cloud business tools (i.e. GDrive, Slack, Jira). I made two major contributions to the company: (1) built the Confluence integration (2) designed, implemented, and led the Historical Scan Platform (HSP) from inception to customer release. Nightfall's core product scans the real-time edits that are made in these tools, and alerts on the detected sensitive info. HSP is an adjacent product to this main product, that scans all the historical data in an app at once. Previously, one-off solutions per app / integration were developed that did not consider serious performance, reliability, or scaling concerns. Alternatively, the distributed system I architected and implemented met robust requirements. It is able to scan data in any application, is highly fault-tolerant (allowing for scans that run for weeks with intermittent failures), can handle massive scale with any number of simultaneous scans, and properly balances the detection model load with real-time scanning concurrently. This project took over a year from initial concept to deployment for large clients, such as DoorDash and Snyk.
I participated in the Spring 2021 cohort of the Georgetown Startup Accelerator with a good friend who is a Georgetown Alum. We pitched the idea for Content Rooms; a digital media platform where creators could post their content (videos, pictures, text, chat-rooms, etc.) individually or with groups of creators, and monetize it via subscriptions instead of advertising. We built a prototype software platform on AWS that could live-stream video, and post text, pictures, and videos behind pay-walls. We did not receive venture capital investment, but the accelerator was an excellent learning experience in building a product from scratch and seeking capitalization.
Wunderkind is an advertising software start-up that sends targeted marketing content over SMS/Email to users of its e-commerce clients. This is accomplish this by probing the client’s e-commerce site to track events that occur during a shopping session. I worked as a backend generalist performing an eclectic set of tasks, which gave me a robust overview of distributed systems and cloud technologies. Some tasks I completed were to refactor core logic of Go micro-services, build REST/gRPC APIs, enrich the event stream buffered in message queues, implement rate limiting, refactor data stores, and improve the monitoring/alerting chain.
I worked on Google’s global-scale distributed time-series database / monitoring system called "Monarch". This system is analogous to DataDog or Prometheus for internal use, and is also the foundation for monitoring and alerting in GCP as well (called "StackDriver"). I worked in the query engine team, where we were specifically focused on optimizing queries and query operators. The project I worked on optimized the implementation of a query operator that took in a set of time-series data streams and output an aggregation of the streams, using a new postulated algorithm. I achieved a substantial performance improvement in almost all cases. This implementation is currently deployed and has likely saved Google hundred of thousands / millions in compute/energy costs to date.