I am concerned with the performance of software on single machines, that is, how do we get the most out of the high-capability computers that are available to us. This work is part systems programming, dev tools and good old fashioned grinding. My primary area of focus is creating a system that makes performance regression detection straightforward for non-experts, to the point where it becomes a CI step, part and parcel of my belief that if you give people good, reliable information they'll steer themselves.

I focus on improving the resource consumption and reliability of the Datadog Agent.

I was the Infrastructure Lead at Goodwater Capital. Infrastructure is responsible for the following, in no particular order:

• Provide a stable and comprehensible deployment tools for our engineering staff.

• Provide a deployment environment that allows our engineering staff to build high-quality software for business users.

• Provide a continuous integration environment that allows our engineering staff to build high-quality software for business users.

• Ensure the security of our compute resources.

• Systematically address instability and insecurity of our existing and future compute resources.

• Partner with engineering staff to debug serious outages.

• Partner with engineering staff to ensure that proposed software architectures meet stability and quality goals.

• Reduce compute costs through optimization of existing systems or redesign.

Infrastructure at Goodwater Capital does not:

• Build features for business end-users.

• Perform quality assurance testing.

• Ensure the stability or quality of end-user focused software independent of the original engineering staff.

In my role as lead I made sure that each of these tracks of responsibility kept ticking over as close to invisibly as possible.

I worked in the Application Performance team, tasked with caring for performance in Dropbox as a whole, addressing regressions in the same through tooling, process updates and direct engineering.

From November 2017 to May 2018 I wrote "Hands-On Concurrency with Rust". This work is an overview of modern hardware concurrency primitives, a motivated tour through Rust standard library internals and important crates. It is my intention that readers finish the book able to build locking primitives from first principles on x86 and ARM, have comfort reading through the Rust standard library code and can link Rust into existing code bases.

Please find a longer post on the writing of the book here: https://blog.troutwine.us/2018/06/07/hands-on-concurrency-with-rust/