Our favorite work is performance optimization. If you need something to run faster, we'll make it faster. Guaranteed.

Anyone can use a line profiler. When doing performance optimization, we look at the problem at its most fundamental, and question every decision along the way. What algorithms are being used? Are they optimal? Required for the common case? What can be effectively cached? Are there resource bottlenecks? Where could custom code provide better solutions over standard implementations?

A few of our recent case studies:

• SPATS Sequence alignment: Custom sequence-alignment algorithms that brought over 1000x performance improvement over industry-standard tools.
• AD Simulation: We've more-than-doubled the framerate of a client's autonomous driving (AD) simulator, while at the same time increasing simulator accuracy and creating much simpler client APIs.

• Matrix-vector products: In support of cryptographic algorithms, we've created code that performs the largest matrix-vector dot products in the world. (Well, okay, we're pretty sure they're the largest...)
• Blockchain Scaling: In the process of developing a proprietary blockchain, we built out a test infrastructure to validate that the blockchain and server code could scale up to millions of users.